WO2026019779A1

WO2026019779A1 - Microfluidic device for recovering rare biological cells from a sample

Info

Publication number: WO2026019779A1
Application number: PCT/US2025/037674
Authority: WO
Inventors: Samuel Z. Williams; Hemant SURYAWANSHI
Original assignee: Columbia University in the City of New York
Current assignee: Columbia University in the City of New York
Priority date: 2024-07-17
Filing date: 2025-07-15
Publication date: 2026-01-22
Anticipated expiration: 2027-01-17

Abstract

Sperm cells can be extracted from a sample by passing the sample through a microfluidic device, capturing images of the sample at a first portion of the microfluidic device, analyzing the images to ascertain whether a sperm cell is present in any given portion of the sample, and using a second portion of the microfluidic device to route only those portions of the sample that include a sperm cell into a container. Notably, the fluid-flow channels in the first portion of the microfluidic device have rectangular roofs, which improves the quality of the captured images. On the other hand, the fluid-flow channels in the second portion of the microfluidic device have curved roofs, which improves the operation of the valves that are used to carry out the selective-routing functions.

Description

MICROFLUIDIC DEVICE FOR RECOVERING RARE BIOLOGICAL CELLS FROM A SAMPLE

CROSS REFERENCE TO RELATED APPLICATIONS

[0001] This Application claims the benefit of US Provisional Applications 63/672,694 (filed July 17, 2024) and 63/794,634 (filed April 25, 2025), each of which is incorporated herein by reference in its entirety.

BACKGROUND

[0002] A normal sperm count is between 15 million to more than 200 million sperm per ml of semen. Below those levels, as long as the semen contains a significant amount of sperm (e.g., around 1 million sperm per ml), many techniques exist for extracting sperm for in vitro fertilization. But some men have extremely low sperm counts, e.g., less than 10 sperm cells per ml. Identification and recovery of sperm cells in semen samples from these men and in testicular biopsy samples has heretofore been challenging because the semen sample includes very few sperm cells but a high number of other cells and debris. Identification and recovery of sperm cells for use in fertilization is also difficult because, in order to use the sperm cells in subsequent procedures, the identification and recovery processes must not damage the sperm cells. The aforementioned challenges also apply to identifying and recovering other types of rare biological cells (e.g., cancer cells, fetal cells, placenta cells, etc.) from within a sample.

[0003] Existing cell identification and sorting technologies do not provide any good solutions for the problems described above. For example, a fluorescence-activated cell sorting (FACS) sorting approach is impractical for sorting sperm cells and other rare cells due to lack of specific monoclonal antibodies against the cells. Further, the high-pressure system used in a FACS machine could potentially damage the cells, and would therefore be detrimental to the use of the cells for any sensitive biological applications. In addition, FACS requires at least 10,000 cells. Other sorting approaches, such as magnetic- activated cell sorting (MACS) based separation, are unsuitable for use with sperm cells that are present in very low numbers or other rare cells because these approaches lack sensitivity and work only at high cell concentration ranges. SUMMARY OF THE INVENTION

[0004] One aspect of this application is directed to a first method of building a microfluidic device. The first method comprises forming a main raised trace with a rectangular cross-section on a first substrate, the main raised trace having a beginning, an end, and a first portion located between the beginning and the end; and forming a first raised trace with a rectangular cross-section on the first substrate. The first raised trace meets the main raised trace between the first portion and the end. The first method also comprises forming a second raised trace with a rectangular cross-section on the first substrate wherein the second raised trace meets the main raised trace between the first portion and the end. The first portion of the main raised trace is located on a first part of the first substrate, and at least part of the first raised trace and at least part of the second raised trace are located on a second part of the first substrate. The first method also comprises heating the second part of the first substrate until the cross-section of at least part of the first raised trace and the cross-section of at least part of the second raised trace become rounded, while ensuring that the first portion of the main trace is not heated to a point where its cross section becomes rounded.

Subsequent to the heating, an uncured polymer material is poured on top of the first substrate, the main raised trace, the first raised trace, and the second raised trace. Subsequent to the pouring, the polymer material is cured to form a transparent flow layer.

[0005] In some instances of the first method, the main raised trace, the first raised trace, and the second raised trace are made from a cured photoresist material. In some instances of the first method, the main raised trace, the first raised trace, and the second raised trace are made from cured AZ 40XT-1 ID photoresist material, and the first substrate comprises a silicon wafer. In some instances of the first method, the first portion of the main raised trace is located at least 1 cm away from the first raised trace and the second raised trace.

[0006] Some instances of the first method further comprise aligning the flow layer to a control layer so that a first control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the first raised trace, and so that a second control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the second raised trace; and bonding the flow layer to the control layer subsequent to the aligning. Optionally, these instances can further comprise bonding the control layer to a transparent slide. [0007] Some instances of the first method further comprise aligning the flow layer to a control layer so that a first control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the first raised trace, and so that a second control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the second raised trace; and bonding the flow layer to the control layer subsequent to the aligning. These instances also comprise, prior to the aligning, forming the control layer by

(a) forming a plurality of raised traces with rectangular cross-sections on a second substrate,

(b) pouring an uncured elastomer material on top of the second substrate and the plurality of raised traces and subsequently curing the elastomer to form the control layer.

[0008] Another aspect of this application is directed to a first microfluidic device that comprises a flow layer and a control layer. The flow layer has a front surface and a rear surface. The rear surface of the flow layer has a main indentation that is shaped and dimensioned to form a main fluid-flow channel having an inlet, the main fluid-flow channel runs along the rear surface of the flow layer, and at least a first portion of the main fluid-flow channel has a flat front surface. The rear surface of the flow layer has a first indentation that is shaped and dimensioned to form a first fluid-flow channel that runs along the rear surface of the flow layer between (a) a portion of the main fluid-flow channel that is downstream from the first portion and (b) a first outlet. At least part of the first fluid-flow channel has a rounded front surface. The control layer is bonded against the rear surface of the flow layer. The control layer has a first control surface that is aligned with the part of the first fluid-flow channel that has a rounded front surface. And the control layer has a first control channel aligned with the first control surface such that pressurizing the first control channel above a first pressure threshold causes the first control surface to bulge into the part of the first fluidflow channel that has a rounded front surface, which prevents fluid from flowing through the first fluid-flow channel.

[0009] In some embodiments of the first microfluidic device, the entire flow layer is formed from a single piece of an elastomer material. Some embodiments of the first microfluidic device further comprise a flat substrate bonded to a rear surface of the control layer.

[0010] In some embodiments of the first microfluidic device, the rear surface of the flow layer has a second indentation that is shaped and dimensioned to form a second fluidflow channel that runs along the rear surface of the flow layer between (a) the portion of the main fluid-flow channel that is downstream from the first portion and (b) a second outlet. At least a part of the second fluid-flow channel has a rounded front surface. In these embodiments, the control layer has a second control surface that is aligned with the part of the second fluid-flow channel that has a rounded front surface. The control layer has a second control channel aligned with the second control surface such that pressurizing the second control channel above a second pressure threshold causes the second control surface to bulge into the part of the second fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel.

[0011] Optionally, in the embodiments described in the previous paragraph, the entire flow layer is formed from a single piece of an elastomer material, and the control layer comprises an elastomer material. Optionally, in the embodiments described in the previous paragraph, the entire flow layer is formed from a single piece of PDMS, and the control layer comprises PDMS. Optionally, the embodiments described in the previous paragraph can further comprise a flat substrate bonded to a rear surface of the control layer.

[0012] Another aspect of this application is directed to a second microfluidic device that includes a substrate, a layer of elastomer material, and a second layer of material. The substrate has a front surface. The layer of elastomer material has a front surface and a rear surface, and the rear surface of the layer of elastomer material is bonded to the front surface of the substrate. The second layer of material has a front surface and a rear surface, and the rear surface of the second layer of material is bonded to the front surface of the layer of elastomer material. The rear surface of the second layer of material has a main indentation that is shaped and dimensioned to form a main fluid-flow channel having an inlet, and the main fluid-flow channel runs along the rear surface of the second layer of material. At least a first portion of the main fluid-flow channel has a flat front surface. The rear surface of the second layer of material has a first indentation that is shaped and dimensioned to form a first fluid-flow channel that runs along the rear surface of the second layer of material between (a) a portion of the main fluid-flow channel that is downstream from the first portion and (b) a first outlet. At least part of the first fluid-flow channel has a rounded front surface. The rear surface of the second layer of material has a second indentation that is shaped and dimensioned to form a second fluid-flow channel that runs along the rear surface of the second layer of material between (a) the portion of the main fluid-flow channel that is downstream from the first portion and (b) a second outlet. And at least a part of the second fluid-flow channel has a rounded front surface. The rear surface of the layer of elastomer material has a first indentation that forms a first control channel, and the first control channel is shaped, dimensioned, and aligned with the first fluid-flow channel so that pressurizing the first control channel above a first pressure threshold causes a first part of the front surface of the layer of elastomer material to bulge forward into the part of the first fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the first fluidflow channel. The rear surface of the layer of elastomer material has a second indentation that forms a second control channel. The second control channel is shaped, dimensioned, and aligned with the second fluid-flow channel so that pressurizing the second control channel above a second pressure threshold causes a second part of the front surface of the layer of elastomer material to bulge forward into the part of the second fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel.

[0013] In some embodiments of the second microfluidic device, the second layer of material comprises an elastomer material. In some embodiments of the second microfluidic device, the layer of elastomer material comprises PDMS, and the second layer of material comprises PDMS.

[0014] Another aspect of this application is directed to a third microfluidic device that comprises a flow layer having a plurality of fluid-flow channels. Each of the fluid-flow channels has a respective roof. Regions of the fluid- flow channels that are used to implement imaging have rectangular roofs, and regions of the fluid-flow channels that are used to implement valve functions have curved roofs.

[0015] Another aspect of this application is directed to a second method of building a microfluidic device. The second method comprises forming a main raised trace with a rectangular cross-section on a first substrate, the main raised trace having a first portion and a final portion; and forming a first raised trace with a rectangular cross-section on the first substrate. The first raised trace meets the final portion of the main raised trace. The second method also comprises forming a second raised trace with a rectangular cross-section on the first substrate. The second raised trace meets the final portion of the main raised trace, the first portion of the main raised trace is located on a first part of the first substrate, and at least part of the first raised trace and at least part of the second raised trace are located on a second part of the first substrate. The second method also comprises heating the second part of the first substrate until the cross-section of at least part of the first raised trace and the crosssection of at least part of the second raised trace become rounded, while ensuring that the first portion of the main trace is not heated to a point where its cross section becomes rounded. The second method also comprises, subsequent to the heating, pouring an uncured polymer material on top of the first substrate, the main raised trace, the first raised trace, and the second raised trace. And the second method also comprises, subsequent to the pouring, curing the polymer material to form a transparent flow layer.

[0016] In some instances of the second method, the main raised trace, the first raised trace, and the second raised trace are made from a cured photoresist material. In some instances of the second method, the main raised trace, the first raised trace, and the second raised trace are made from cured AZ 40XT-11D photoresist material, and the first substrate comprises a silicon wafer. In some instances of the second method, the first portion of the main raised trace is located at least 1 cm away from the first raised trace and the second raised trace.

[0017] Some instances of the second method further comprise aligning the flow layer to a control layer so that a first control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the first raised trace, and so that a second control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the second raised trace; and bonding the flow layer to the control layer subsequent to the aligning. Optionally, these instances can further comprise bonding the control layer to a transparent slide.

[0018] Some instances of the second method further comprise aligning the flow layer to a control layer so that a first control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the first raised trace, and so that a second control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the second raised trace; and bonding the flow layer to the control layer subsequent to the aligning. These instances also further comprise, prior to the aligning, forming the control layer by (a) forming a plurality of raised traces with rectangular crosssections on a second substrate, and (b) pouring an uncured elastomer material on top of the second substrate and the plurality of raised traces and subsequently curing the elastomer to form the control layer. [0019] Another aspect of this application is directed to a fourth microfluidic device. The fourth microfluidic device comprises a flow layer and a control layer. The flow layer has a front surface and a rear surface. The rear surface of the flow layer has a main indentation that is shaped and dimensioned to form a main fluid-flow channel having an inlet, a first portion having a flat front surface, and a final portion that is downstream from the first portion. The main fluid-flow channel runs along the rear surface of the flow layer. The rear surface of the flow layer has a first indentation that is shaped and dimensioned to form a first fluid-flow channel that runs along the rear surface of the flow layer between (a) the final portion of the main fluid-flow channel and (b) a first outlet. At least part of the first fluidflow channel has a rounded front surface. The control layer is bonded against the rear surface of the flow layer. The control layer has a first control surface that is aligned with the part of the first fluid-flow channel that has a rounded front surface. And the control layer has a first control channel aligned with the first control surface such that pressurizing the first control channel above a first pressure threshold causes the first control surface to bulge into the part of the first fluid- flow channel that has a rounded front surface, which prevents fluid from flowing through the first fluid-flow channel.

[0020] In some embodiments of the fourth microfluidic device, the first portion and the final portion of the main fluid-flow channel are both straight, and the first portion of the main fluid-flow channel empties directly into the final portion of the main fluid-flow channel. In some embodiments of the fourth microfluidic device, the main indentation is shaped so that the main fluid- flow channel has a serpentine second portion disposed between the first portion and the final portion.

[0021] In some embodiments of the fourth microfluidic device, the final portion of the main fluid-flow channel has a front surface that transitions smoothly from flat to rounded. In some embodiments of the fourth microfluidic device, the entire flow layer is formed from a single piece of an elastomer material. Some embodiments of the fourth microfluidic device further comprise a flat substrate bonded to a rear surface of the control layer.

[0022] In some embodiments of the fourth microfluidic device, the rear surface of the flow layer has a second indentation that is shaped and dimensioned to form a second fluidflow channel that runs along the rear surface of the flow layer between (a) the final portion of the main fluid-flow channel and (b) a second outlet. At least a part of the second fluid-flow channel has a rounded front surface. The control layer has a second control surface that is aligned with the part of the second fluid- flow channel that has a rounded front surface, and the control layer also has a second control channel aligned with the second control surface such that pressurizing the second control channel above a second pressure threshold causes the second control surface to bulge into the part of the second fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel.

[0023] Optionally, in the embodiments described in the previous paragraph, the entire flow layer can be formed from a single piece of an elastomer material, and the control layer comprises an elastomer material. Optionally, in the embodiments described in the previous paragraph, the entire flow layer is formed from a single piece of PDMS, and the control layer comprises PDMS. Optionally, the embodiments described in the previous paragraph can further comprise a flat substrate bonded to a rear surface of the control layer.

[0024] Another aspect of this application is directed to a fifth microfluidic device. The fifth microfluidic device comprises a substrate having a front surface, a layer of elastomer material, and a second layer of material. The layer of elastomer material has a front surface and a rear surface, and the rear surface of the layer of elastomer material is bonded to the front surface of the substrate. The second layer of material has a front surface and a rear surface, and the rear surface of the second layer of material is bonded to the front surface of the layer of elastomer material. The rear surface of the second layer of material has a main indentation that is shaped and dimensioned to form a main fluid-flow channel having an inlet, a first portion having a flat front surface, and a final portion that is downstream from the first portion, and the main fluid- flow channel runs along the rear surface of the second layer of material. The rear surface of the second layer of material has a first indentation that is shaped and dimensioned to form a first fluid-flow channel that runs along the rear surface of the second layer of material between (a) the final portion of the main fluid-flow channel and (b) a first outlet. At least part of the first fluid-flow channel has a rounded front surface. The rear surface of the second layer of material has a second indentation that is shaped and dimensioned to form a second fluid-flow channel that runs along the rear surface of the second layer of material between (a) the final portion of the main fluid-flow channel and (b) a second outlet. At least a part of the second fluid- flow channel has a rounded front surface. The rear surface of the layer of elastomer material has a first indentation that forms a first control channel. The first control channel is shaped, dimensioned, and aligned with the first fluid-flow channel so that pressurizing the first control channel above a first pressure threshold causes a first part of the front surface of the layer of elastomer material to bulge forward into the part of the first fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the first fluid-flow channel. The rear surface of the layer of elastomer material has a second indentation that forms a second control channel. And the second control channel is shaped, dimensioned, and aligned with the second fluid-flow channel so that pressurizing the second control channel above a second pressure threshold causes a second part of the front surface of the layer of elastomer material to bulge forward into the part of the second fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel.

[0025] In some embodiments of the fifth microfluidic device, the first portion and the final portion of the main fluid-flow channel are both straight, and the first portion of the main fluid-flow channel empties directly into the final portion of the main fluid-flow channel. In some embodiments of the fifth microfluidic device, the main indentation is shaped so that the main fluid-flow channel has a serpentine second portion disposed between the first portion and the final portion.

[0026] In some embodiments of the fifth microfluidic device, the final portion of the main fluid-flow channel has a front surface that transitions smoothly from flat to rounded. In some embodiments of the fifth microfluidic device, the second layer of material comprises an elastomer material. In some embodiments of the fifth microfluidic device, the layer of elastomer material comprises PDMS, and the second layer of material comprises PDMS.

[0027] Another aspect of this application is directed to a sixth microfluidic device. The sixth microfluidic device comprises a flow layer having a front surface and a rear surface, and a control layer bonded against the rear surface of the flow layer. The flow layer has a plurality of fluid-flow channels, each of which has a respective roof. Regions of the fluid-flow channels that are used to implement imaging have rectangular roofs, and regions of the fluid-flow channels that are used to implement valve functions have curved roofs.

BRIEF DESCRIPTION OF THE DRAWINGS

[0028] FIG. 1 illustrates an embodiment of a system for identifying and recovering target cells present in a sample. [0029] FIGS. 2A and 2B respectively depict plan views of the fluid- flow layer and the control layer of a microfluidics device.

[0030] FIG. 2C depicts the microfluidics device that results from combining the fluidflow layer and the control layer.

[0031] FIG. 3 A shows the positions of the various layers of the microfluidic device of FIG. 2C when a channel of the control layer is not pressurized.

[0032] FIG. 3B shows the positions of the various layers of the microfluidic device of FIG. 2C when a channel of the control layer is pressurized.

[0033] FIG. 4A and 4B respectively depict images of a sample flowing through a fluid-flow channel with a rectangular roof and a curved roof, respectively.

[0034] FIGS. 5A-5D depict a process for building the microfluidic device of FIG. 2C.

[0035] FIGS. 6A-6C depicts another microfluidics device that is similar to the FIGS. 2A-C device, except that the main fluid-flow channel is straight.

[0036] FIG. 7 depicts a process step for building the microfluidic device of FIG. 6C.

[0037] Various embodiments are described in detail below with reference to the accompanying drawings, wherein like reference numerals represent like elements.

DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0038] In view of the shortcomings of the prior art identified above, it would be desirable to provide a system and method that are capable of accurately and efficiently identifying and capturing sperm cells from samples that contain an extremely low number of sperm cells, without damaging those sperm cells, so that they can subsequently be used (e.g., for in vitro fertilization).

[0039] Ordinarily, if a semen sample that contains very few sperm cells (e.g., less than 100 sperm cells per ml) is pumped through a channel that is large enough for semen to flow through without clogging, conventional techniques will not be able to identify and capture the sperm cells. Two factors contribute to this difficulty: first, other cells and debris in the semen can outnumber the target sperm cells by over 100,000: 1. And second, the cross- section of the channel is so large that it is impossible to keep the entire cross-section of the channel in focus to identify the extremely rare target sperm cells for subsequent capture.

[0040] The embodiments described herein overcome this problem by, inter alia, making sure that non-turbulent flow (e.g., laminar flow) is maintained through the channel. When laminar flow is maintained, the flow is slower near the walls of the channel and faster near the center of the channel. As a result, the majority of the sample ends up flowing through the center of the channel. Because the majority of the sample flows through the center of the channel, it becomes possible for a microscope to maintain focus on the entire cross-section of the center of the channel through which the majority of the sample flows. And when the entire cross-section of flow can be kept in focus, the system can spot the rare sperm cells in images captured using the microscope and a camera as they traverse the channel.

[0041] FIG. 1 illustrates an embodiment of a system for identifying and recovering target cells present in a sample. The system will be described in the context of identifying and recovering sperm cells present in a semen sample, but can also be used in a variety of other contexts.

[0042] The FIG. 1 embodiment includes a syringe pump 12, a microfluidics device 20 that includes a main fluid-flow channel 35, a camera 50, and a processor 70. The syringe pump 12 pumps a semen sample containing sperm cells through the main fluid-flow channel 35 of the microfluidics device 20. In alternative embodiments, a different type of pump can be used instead of the illustrated syringe pump 12. The camera 50 captures images of the semen sample and the sperm cells therein as the semen sample flows through a first portion 35F (shown in FIG. 2A) of the main fluid-flow channel 35. Thus, the microfluidics device 20 should be transparent, at least in the vicinity of the first portion 35F. And the processor 70 detects which of the captured images include a sperm cell.

[0043] The microfluidics device 20 includes a set of valves that are actuated by a valve controller 80 that is controlled by the processor 70. In the embodiment depicted in FIG. 1, the valve controller relies on a compressed air cylinder as a source of air pressure. But in alternative embodiments, other sources of compressed air (or another gas) can be used, e.g., an air compressor pump. As described below in connection with FIG. 2A-2C, the microfluidics device 20 includes two valves, and the presence of air pressure in a corresponding control channel will close the corresponding valve to divert the flow in the desired direction. The air pressure that is used to control the valves could be, for example, 5- 10 psi.

[0044] Ordinarily, the processor 70 instructs the valve controller 80 to set the state of the set of valves so that the fluid that exits the main fluid- flow channel 35 will flow out via outlet 1. But if the processor 70 detects the presence of a sperm cell, the processor determines when that sperm cell will exit the main fluid- flow channel 35. (This may be accomplished, for example, based on knowledge of the flow rate and the distance between the point at which the sperm cell was detected and the end of the main fluid-flow channel 35.) The processor 70 then sends signals to the valve controller 80 to set the state of the valves so that just before the sperm cell exits the main fluid-flow channel 35, the valves will be set so that the exiting fluid (which includes the sperm cell) will flow out via outlet 2. A short time thereafter (e.g., 1-15 seconds, 5-15 seconds, 10-15 seconds, 2-5 seconds, 5-10 seconds, etc.), the processor 70 sends signals to the valve controller 80 to return the state of the valves to their default setting so that the fluid that exits the main fluid-flow channel 35 will flow out via outlet 1.

[0045] To ensure that the camera 50 is able to capture sharply focused images and the sample collection station properly captures portions of the semen sample, it is important for the semen sample to have a consistent, precise flow rate through the main fluid- flow channel 35 of the microfluidics device 20. A semen sample will typically have a volume of 1 to 3 ml. In the illustrated embodiment, a syringe pump 12 is configured to pump the semen sample through the main fluid-flow channel 35. But in alternative embodiments, different types of pumps may be used. In some embodiments, the pump 12 pumps the semen sample through the main fluid-flow channel 35 at a flow rate of 200-2000 pL/hr (e.g., 300-1000 pL/hr). The inventors have experimentally determined that flow rates in these ranges can advantageously process human semen samples quickly without causing image blurring. But in alternative embodiments, flow rates outside these ranges can also be used.

[0046] If the semen sample forms clumps of semen that do not flow consistently in the first portion 35F of the main fluid-flow channel 35, the camera 50 cannot accurately image the sperm cells, which would prevent the system from tracking and recovering at least some of the sperm cells. And because a short channel height can lead to clumping, the height of the first portion 35F as well as the rest of the main fluid-flow channel 35 must not be too small. [0047] If the sperm cells in the semen sample are stacked or widely distributed through the cross-section of the first portion 35F of the main fluid-flow channel 35 in the direction of the viewing axis of the camera 50 (i.e., the height direction), the camera 50 will fail to image the majority of the sperm cells as they flow through the first portion 35F and/or will capture unfocused images of the majority of the sperm cells. That is, such stacking or distribution of sperm cells will cause the majority of the sperm cells to be disposed outside of the depth of field (DOF) of the camera 50. For this region, the height of the first portion 35F should not be too large.

[0048] Further, unless care is taken to ensure that the flow of the sample through the first portion 35F is non-turbulent, turbulence in the sample will cause blurring in images captured by the camera 50 due to erratic motion of the semen sample, and will prevent the flow of the semen sample from being concentrated at a vertically central portion of the first portion 35F, thereby causing the problem of stacking and wide distribution of the sperm cells along the height of the main fluid- flow channel 35.

[0049] To avoid the problems described above and enable the camera 50 to capture well-focused images of all of the sperm cells in the semen sample, the main fluid- flow channel 35 and the inlet to the main fluid-flow channel 35 are shaped and dimensioned to provide a non-turbulent (e.g., laminar) flow of the semen sample through the main fluid-flow channel 35, and to maintain the flow of the semen sample at a vertically central portion of the main fluid-flow channel 35, especially at the first portion 35F of the main fluid-flow channel 35, which is where images are captured by the microscope.

[0050] One example of suitable dimensions to facilitate non-turbulent, laminar, flow in a substantially rectangular main fluid-flow channel 35, is a height of 40 pm, and a width of 300 pm. In some embodiments, the inlet to the main fluid-flow channel 35 has a shape that includes a conical transition that narrows as the fluid flows into the inlet. These dimensions for the main fluid- flow channel 35 provide laminar flow of the semen samples through the main fluid-flow channel 35 without clumping, with the flow of the semen samples located within a vertically central portion of the main fluid- flow channel 35.

[0051] Notably, with the foregoing configurations of the microfluidics device 20, most of the sample will flow with a laminar flow through the central-most 8 pm portion of the first portion 35F of the main fluid-flow channel 35. In some embodiments, most of the sample will flow through the central-most 6 pm portion, the central-most 5 pm portion, the central-most 4 pm portion, the central-most 3 pm portion, or even the central-most 2 pm portion of the first portion 35F. Thus, the flow of the sample is confined within a vertically central region of the first portion 35F, such that the target cells are not vertically stacked in the first portion 35F and the flow of the sample falls within a DOF of the camera 50, which can be small (e.g., 2, 3, 4, 5, 6, or 8 pm).

[0052] In alternative embodiments, the height of the main fluid-flow channel 35 could range from 10 to 50 pm, and the width of the main fluid- flow channel 35 could range from 100 to 400 pm. The length of the main fluid-flow channel 35 is not critical.

[0053] Maintaining non-turbulent (e.g., laminar) flow is important because turbulent flow would cause the sperm cells and debris to be distributed across the full z-axis, which would make it difficult if not impossible to keep the relevant cells in focus and to keep track of the sperm cells as they traverse the flow channel. For if the sperm cells were distributed across the full z-axis, only a small fraction of sperm would be visible in any given image. In contrast, when non-turbulent (e.g., laminar) flow is maintained, the majority of the sample (including the sperm cells) will flow through a vertically central region of the first portion 35F of the main fluid-flow channel 35 that is short enough in height (e.g., a few pm) so that the majority of the sample (including the sperm cells) will remain in focus as they traverse the first portion 35F. This is important because it allows the sperm cells to be visualized by the camera 50 as they pass by the field-of-view of the microscope.

[0054] The microfluidics device 20 is positioned on a microscope (not shown) such that the first portion 35F of the main fluid-flow channel 35 is optically aligned with an objective lens 55 of the microscope. The microscope can be a phase contrast microscope, but is not limited thereto.

[0055] The camera 50 includes one or more lenses configured to be aligned with the objective lens 55 to capture images of the semen sample as the semen sample flows through the first portion 35F of the main fluid-flow channel 35. Since the flow of a semen sample through the first portion 35F is confined to the vertically central portion of the first portion 35F, as described above, the camera 50 can be precisely focused on the entire vertically central portion of the first portion 35F and can therefore capture clear, properly focused images of most of the sperm cells in the semen sample. For example, the camera 50 can be focused on a vertically central portion of the first portion 35F having a height in a range of 2 to 8 pm.

[0056] The camera 50 is a high-speed camera. In some embodiments, the resolution, shutter speed, frame rate, trigger time, and other parameters of the camera 50 are such that the camera acquires multiple images of sperm cells in the semen sample as the sperm cells pass through the first portion 35F of the main fluid-flow channel 35. And the parameters of the camera 50 are optimized to make sure captured images are not blurry and the same sperm cell is captured in multiple images, at various locations along the first portion 35F of the main fluid-flow channel 35. The camera 50 can be configured to capture images at a frame rate of at least 100 frames per second. For example, the camera 50 can capture images at a frame rate of 300-500 frames per second. The camera 50 can capture images of the same sperm cell many time (e.g., at least 5 times) as the sperm cell passes through the first portion 35F of the main fluid-flow channel 35. Capturing multiple images of the same sperm cell as the sperm cell travels along the first portion 35F can advantageously improve the system’s confidence that what appears to be a sperm cell in one image is indeed a sperm cell. But in alternative embodiments, when sufficiently powerful image recognition techniques are available, the system can base its decision as to whether a sperm cell is present based on only a single image.

[0057] Operation of the camera 50 can be controlled by the processor 70, or by a dedicated controller of the camera 50. Images captured by the camera 50 can be stored in a memory, which can include one or more high-capacity solid state drives (SSDs). The processor 70 analyzes the images captured by the camera 50 to identify target sperm cells in the semen sample and keeps track of which images contain the target sperm cells (e.g., in a memory, not shown).

[0058] The processor 70 inputs each of the captured images of the semen sample flowing through the first portion 35F of the main fluid-flow channel 35, and identifies which images contain a sperm cell. This identification can be accomplished, for example, by having the processor 70 implement one or more GPU-based computational systems that apply a pretrained neural network to the images captured by the camera 50 to identify the target sperm cells in the images. Alternatively, the processor 70 can be programmed to implement an object detection model. For example, the object detection model can be a trained “you only look once” (YOLO) object detection model (e.g., YOLOv4). The trained object detection model can be trained based on many (e.g., millions) of training images containing sperm cells being input to an object detection model in a training phase.

[0059] As explained above, the processor 70 sends signals to the valve controller 80 to set the state of the valves so that portions of the sample that do not include a sperm cell will flow out via outlet 1, and so that portions of the sample that do include a sperm cell will flow out via outlet 2.

[0060] FIG. 2C depicts a plan view of the microfluidics device 20; FIGS. 2A and 2B respectively depict plan views of the fluid- flow layer 30 and the control layer 40 of that microfluidics device 20; and FIGS. 3A-B are side views of the microfluidics device 20. This microfluidics device 20 is optimized to enable high-throughput imaging in the region of interest (ROI), and successful recovery of the sperm cells. The microfluidics device 20 is formed by using two PDMS chips that are bonded together. The upper/front PDMS layer depicted in FIGS. 2A and 3A-B is the flow layer which includes the channel through which the sample flows. The lower/rear PDMS layer depicted in FIGS. 2B and 3A-B is the control layer 40. Each of these layers 30, 40 can be, e.g., 20-40 pm thick. The flow layer 30 and the control layer 40 are aligned as depicted in FIGS. 2C and 3A-B so that the presence or absence of air pressure in each channel of the control layer 40 will either close or open a corresponding flow path in the flow layer 30. A clear flat substrate 45 (e.g., a glass slide) is bonded to the bottom of the lower PDMS layer, as best seen in FIGS. 3A-B. In alternative embodiments, the flow layer 30 can be made from a clear material other than PDMS (e.g., glass), and the control layer 40 can be made from elastomers other than PDMS.

[0061] The flow layer 30 has a front surface and a rear surface. The rear surface of the flow layer has a main indentation that is shaped and dimensioned to form a main fluidflow channel 35 having an inlet, a first portion 35F having a flat front surface, and a final portion 35N that is downstream from the first portion. As best seen in FIGS. 2A and 2C, the main fluid-flow channel 35 in this embodiment 20 also has a serpentine second portion 35S disposed between the first portion 35F and the final portion 35N. The various sections of the main fluid-flow channel 35 are arranged so that the sample enters the sample inlet, then sequentially flows through the first portion 35F, the second portion 35S, and the final portion 35N. Optionally, an additional initial portion (not shown) can sit between the inlet and the first portion 35F of the main fluid-flow channel 35. [0062] The main fluid-flow channel 35 runs along the rear surface of the flow layer 30. The rear surface of the flow layer 30 also has a first indentation that is shaped and dimensioned to form a first fluid-flow channel 1 that runs along the rear surface of the flow layer between (a) the final portion 35N of the main fluid- flow channel and (b) a first outlet 1. And the rear surface of the flow layer 30 also has a second indentation that is shaped and dimensioned to form a second fluid-flow channel 32 that runs along the rear surface of the flow layer 30 between (a) the final portion 35N of the main fluid-flow channel and (b) a second outlet 2. At least part of the first fluid-flow channel 31 has a rounded front surface, and at least a part of the second fluid-flow channel 32 has a rounded front surface. In some embodiments, the flow layer 30 is formed from a single piece of an elastomer material e.g., PDMS.

[0063] The control layer 40 is bonded against the rear surface of the flow layer 30. The control layer has a first control surface that is aligned with the part of the first fluid-flow channel 31 that has a rounded front surface, and a second control surface that is aligned with the part of the second fluid-flow channel 32 that has a rounded front surface. The control layer 40 can be made of PDMS or another elastomer material.

[0064] The control layer 40 has a first control channel 41 aligned with the first control surface such that pressurizing the first control channel 41 above a first pressure threshold causes the first control surface to bulge into the part of the first fluid-flow channel 31 that has a rounded front surface, which prevents fluid from flowing through the first fluid- flow channel 31. The control layer 40 also has a second control channel 42 aligned with the second control surface such that pressurizing the second control channel 42 above a second pressure threshold causes the second control surface to bulge into the part of the second fluid-flow channel 32 that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel 32.

[0065] When the first fluid-flow channel 31 is open and the second fluid-flow channel 32 is closed, fluid in the final portion 35N of the main fluid-flow channel 35 will exit the microfluidics device 20 via the first fluid-flow channel 31 and the first outlet 1. On the other hand, when the first fluid-flow channel 31 is closed and the second fluid-flow channel 32 is open, fluid in the final portion 35N of the main fluid-flow channel 35 will exit the microfluidics device 20 via the second fluid-flow channel 32 and the second outlet 2. [0066] FIGS. 3A-B are side views of the microfluidics device 20 that depict the operation of the valves in the microfluidic device 20. More specifically, FIG. 3A shows the positions of the various layers 30, 40 when a channel of the control layer 40 is not pressurized. In this situation, the control surface on top of the control channel 41/42 will lie flat, in which case the flow channel 31/32 that resides directly above the control surface will remain opened. This corresponds to an open valve. On the other hand, FIG. 3B shows the positions of the various layers when a channel of the control layer 40 is pressurized. In this situation, the control surface on top of the control channel 41/42 will bulge upwards into the flow channel, and will block the flow channel that resides directly above the control surface. This corresponds to a closed valve.

[0067] The operation of the valve depicted in FIG. 3A-B works well when the roof of the flow channel 31/32 in the flow layer 30 is curved. For if the depicted curved roof were to be replaced with a rectangular roof, it would be much more difficult (if not impossible) for the control surface on top of the control channel 41/42 to completely seal off the corresponding sample flow channel 31/32. In other words, using curved roofs to implement the fluid-flow channels in the flow layer 30 of the microfluidic device 20 improves the operation of the valves.

[0068] But using curved roofs to implement the fluid-flow channels in the flow layer 30 of the microfluidic device 20 can cause a very serious problem. This is because the curved roofs distort the images of the sperm cells that are being captured as they travel through the main fluid-flow channel 35 to the point where they become unusable. Compare FIG. 4A (which depicts an image of a sample flowing through a fluid- flow channel with a rectangular roof) to FIG. 4B (which depicts an image of a sample flowing through a fluid-flow channel with a curved roof). In other words, using rectangular roofs to implement the fluid-flow channels in the flow layer 30 of the microfluidic device 20 improves the images.

[0069] Because rectangular roofs improve the image quality, and curved roofs improve the operation of the valves, using a single type of roof for all of the fluid-flow channels will have a negative impact on at least one of (a) valve operation or (b) image quality. The solution is to customize the roof shape in different regions of the microfluidic device 20, so that the roof shape at each region is optimized to the function that will be performed at that region. [0070] FIGS. 5A-D depict one example of a process for building such a microfluidic device 20 using a flow layer 30, a control layer 40, and a substrate (e.g., a glass slide) 45. The initial steps for forming the flow layer 30 from PDMS are depicted in FIG. 5A. The first step in the device fabrication is to produce Si wafer master molds. Fabricating the flow layer master mold involves coating a clean silicon wafer with a uniform layer of positive photoresist via spin coating, followed by a soft bake to enhance adhesion. The desired pattern is then transferred onto the photoresist-coated Si wafer using LJV exposure through a photomask, initiating a chemical reaction in the exposed regions. After a post-exposure bake to finalize the pattern, the exposed photoresist is selectively removed during development, revealing the patterned photoresist layer. To further cure the remaining photoresist and improve durability, a process of hard baking is carried out. The resulting cured photoresist features are rectangular, as best seen on the bottom of FIG. 5 A.

[0071] Notably, many positive photoresists (e.g., AZ 40XT-11D40 undergo thermal softening and roundening when they are heated e.g., to 110-140° C. The trick is to heat only the portions of the photoresist that corresponded to valve functions, without heating the portions of the photoresist that correspond to imaging functions. FIG. 5B shows how this can be accomplished. More specifically, when only the right half of the silicon wafer is placed on a hotplate, the right half of the silicon wafer will heat up first. Because the portions of the flow channel that correspond to valve functions (i.e., the first and second fluid-flow channels 31, 32) are located on the right half of the silicon wafer, the corresponding photoresist traces will become rounded in 1-3 minutes, as depicted in the top right comer of FIG. 5B. In contrast, because the first portion 35F of the main fluid-flow channel (where imaging is performed) is located on the left half of the silicon wafer (e.g., more than 1 cm away from the hotplate, it will not reach 110° C in 3 minutes). As a result, the corresponding photoresist traces will remain rectangular, as depicted in the top left comer of FIG. 5B.

[0072] Furthermore, because only the rightmost part of the photoresist traces that correspond to the final portion 35N of the main fluid-flow channel 35 sit on the hotplate, and because the center of the hotplate can get hotter than the edge of the hotplate, a gradient of heat will be imposed on the photoresist traces that correspond to the final portion 35N. More specifically, in the example depicted in FIG. 5B, the temperature at point X will be higher than the temperature at point Y, which will in turn be higher than the temperature at point Z. This gradient of the heat will cause the photoresist traces to become fully rounded at point X, somewhat less rounded at point Y, and even less rounded at point Z. (Some rounding can occur at point Z due to thermal conductivity.) The result is a gradual transition from rectangular features to rounded features as you move from left to right along the photoresist traces that correspond to the final portion 35N of the main fluid-flow channel 35. Note that heat sources other than hotplates may be used to round the photoresist traces, including but not limited to heat strips and the Omega™ KHRA, KHLVA, and KHA series of self-adhesive poly imide flexible heaters.

[0073] The control layer 40 can be formed (e.g., from PDMS or another elastomer) using any conventional process. For example, a master mold for the control later can be fabricated by spin coating negative photoresist, followed by a soft bake and transferring the desired pattern through a photomask using UV exposure, as depicted in FIG. 5C. Next, the mold is developed and hard baked leaving behind rectangle shape features.

[0074] Once master molds or both the flow layer and the control layer are ready, they undergo silanization. This helps in peeling off PDMS in the next step. One suitable approach for fabricating the microfluidic device 20 is depicted in FIG. 5D. For the flow layer, PDMS and a curing agent are mixed in an appropriate ratio (e.g., 10:1), degassed, and poured over the flow layer master mold to form a thick layer. Once the PDMS is cured, it is gently peeled off from the master mold and the inlets and outlets are punched. For the control layer, PDMS and a curing agent are mixed, degassed, and spin coated on the control layer master mold. After curing, the flow layer PDMS that was peeled off and the spin coated PDMS on the control mold are plasma treated.

[0075] The flow layer PDMS 30 is aligned on top of the spin coated control layer PDMS 40 such that the valves are aligned to respective outlet channels. During the bonding steps, air bubbles trapped between the PDMS layers are carefully removed. These two PDMS layers 30, 40, along with the control layer master mold (at the base) are placed on a hotplate for further enhancing the bonding. The bonded PDMS layers 30, 40 are peeled off from the control layer master mold and the inlets for the control layer PDMS 40 are punched. In the final step, the two layers of PDMS (i.e., the flow layer 30 and the control layer 40) are bonded to the glass slide 45.

[0076] Each trace of photoresist on the mold for the control layer will give rise to a corresponding indentation on the bottom of the control layer, and each trace of photoresist on the mold for the flow layer will give rise to a corresponding indentation on the bottom of the flow layer. And because some of the traces on the mold for the flow layer are rounded while other traces are rectangular, some of the resulting indentations on the flow layer will have rounded roofs while other indentations on the flow layer will have rectangular roofs. Notably, because the photoresist traces that were used to form the flow layer 30 gradually transition from rectangular features to rounded features as you move from left to right, the front surface (i.e., the roof) of the final portion 35N of the main fluid-flow channel 35 will also gradually transition from flat to rounded in a corresponding fashion. And this gradual transition helps maintain the non-turbulent nature of the flow within the main fluid-flow channel 35.

[0077] Returning to FIGS. 1 and 2C, the syringe pump 12 pumps the sample into the inlet on the left side of the main fluid- flow channel 35 at a constant flowrate. In alternative embodiments, a different type of pump can be used instead of the illustrated syringe pump 12. Imaging is performed in the first portion 35F of the main fluid-flow channel 35. The long serpentine second portion 35S that follows the first portion 35F provides adequate time (e.g., on the order of 5 seconds) for the processor 70 to analyze all the images and decide if a sperm cell is present or not. By default, the first control channel is not pressurized and the second control channel is pressurized, which opens the outlet 1 valve and closes the outlet 2 valve. This causes the output of the main fluid-flow channel 35 to flow out via outlet 1. When a sperm cell is detected, the processor 70 sends signals to the valve controller 80 to, at the correct time, to pressurize the first control channel and unpressurize the second control channel. This closes the outlet 1 valve and opens the outlet 2 valve, which causes the output of the main fluid-flow channel 35 (which includes the identified sperm cell) to flow out via outlet 2. After an appropriate quantity (e.g., 1-2 pL) of the sample has flowed out of outlet 2, the processor 70 sends signals to the valve controller 80 to return to its original state. The inlets and outlets can be connected with medical grade tubing e.g., with an inner diameter of 250-500 pm (e.g., 380 pm).

[0078] Note that if the image-recognition techniques are fast enough (e.g., fast enough to recognize a sperm cell in 1 second or less), the serpentine second portion 35S depicted in FIG. 2C can be omitted, as depicted in FIG. 6. More specifically, FIG. 6 depicts a microfluidics device 20’ that is similar to the device depicted in FIG. 2C in all respects, except that the serpentine second portion 35S of the main fluid-flow channel 35 in the FIG. 2C embodiment is omitted, and the final portion 35N is somewhat longer. Thus, in this embodiment 30’, (1) the first portion 35F and the final portion 35N of the main fluid-flow channel 35’ are both straight, and (2) the first portion 35F of the main fluid-flow channel 35’ empties directly into the final portion 35N of the main fluid-flow channel 35’.

[0079] The microfluidics device 20’ depicted in FIG. 6C has significant advantages over the FIG. 2C embodiment. First, because the path between the first area 35F (where imaging takes place) and the final portion 35N is significantly shorter and straight, the time when a detected sperm cell will arrive at the end of the final portion 35N can be predicted with greater accuracy. This enables the processor 70 to control the valves so that a smaller portion of the sample (e.g., < 0.5 pL, or 0.2-0.4 pL) will flow out via outlet 2 each time a sperm cell is detected, as opposed to the larger volume (e.g., 1-2 pL) that flows out via outlet 2 during each actuation in the FIG. 2C embodiment. The end result is that the recovered sperm cells will be output in a smaller volume of fluid that has a higher sperm concentration, which makes it easier to identify and recover the sperm cells from the output liquid. Second, the straight path in the FIG. 6C embodiment shortens the transit time of the sperm cells within the main fluid-flow channel 35’, minimizes shear stress on the sperm cells, and reduces the sperm cells’ exposure to microfluidic materials, all of which can increase the sperm cells’ viability. And finally, the streamlined channel design reduces manufacturing complexity and cost, reduces the risk of clogging, and minimizes debris accumulation and dead zones.

[0080] FIG. 7 shows how to obtain a flat roof in the imaging portion 35F of the main fluid-flow channel 35’ and a rounded roof in those portions of the flow layer 30’ that correspond to valve functions in the FIG. 6C embodiment by selectively heating only those portions of the photoresist that correspond to valve functions during the manufacturing process. This is similar to the situation described above in connection with FIG. 5B. More specifically, when only the right half of the silicon wafer is placed on a hotplate (or other heater), the right half of the silicon wafer will heat up first. Because the portions of the flow channel that correspond to valve functions (i.e., the first and second fluid-flow channels 31 , 32) are located on the right half of the silicon wafer, the corresponding photoresist traces will become rounded, as depicted in the top right corner of FIG. 7. In contrast, because the first portion 35F of the main fluid-flow channel (where imaging is performed) is located on the left half of the silicon wafer (e.g., more than 1 cm away from the hotplate, it will not reach 110° C in 3 minutes). As a result, the corresponding photoresist traces will remain rectangular, as depicted in the top left comer of FIG. 7.

[0081] Similar to the situation described above in connection with FIG. 5B, because only the rightmost part of the photoresist traces in FIG. 7 that correspond to the final portion 35N of the main fluid-flow channel sit on the hotplate, and because the center of the hotplate can get hotter than the edge of the hotplate, a gradient of heat will be imposed on the photoresist traces that correspond to the final portion 35N. The result is a gradual transition from rectangular features to rounded features as you move from left to right along the photoresist traces that correspond to the final portion 35N of the main fluid-flow channel. And because the photoresist traces that are used to form the flow layer 30 gradually transition from rectangular features to rounded features as you move from left to right, the front surface (i.e., the roof) of the final portion 35N of the main fluid-flow channel will also gradually transition from flat to rounded in a corresponding fashion. And this gradual transition helps maintain the non-turbulent nature of the flow within the main fluid-flow channel.

[0082] The embodiments disclosed herein can be used to identify and recover sperm cells from a semen sample that has an extremely low sperm count. Moreover, the disclosed systems and methods do not need to employ potentially harmful markers such as stains, lasers, antibodies, or other markers to identify the sperm cells, and do not require the application of mechanical or electrical force to capture the sperm cells. Therefore, the sperm cells can be recovered without being damaged, and thus can be used in subsequent procedures after recovery.

[0083] The embodiments disclosed herein can also be used to identify and recover sperm cells from semen samples that have normal sperm counts or sperm counts that are low, but not extremely low. In these situations, the sperm sample can be diluted with a buffer, and the diluted sperm sample can then be run through any of the embodiments described above.

[0084] The foregoing description describes systems and methods in a context in which a semen sample is the sample to be processed and sperm cells are the target cells to be identified and recovered. However, identifying and recovering other types of rare biological cells in other contexts present similar concerns. Therefore, the systems and methods disclosed herein are not limited to identifying and recovering sperm cells. To the contrary - the disclosed systems and methods can identify and recover other types of target cells from other types of samples. For example, the disclosed systems and methods can identify and recover rare biological cells such as, but not limited to, cancer cells, fetal cells, or placenta cells present in a blood sample, a placental fluid sample, or another fluid sample. Additionally, the disclosed systems and methods can be used to perform a complete blood count (CBC), which is routinely used to detect leukemia, anemia, infections, and other disorders.

[0085] While the present invention has been disclosed with reference to certain embodiments, numerous modifications, alterations, and changes to the described embodiments are possible without departing from the sphere and scope of the present invention, as defined in the appended claims. Accordingly, it is intended that the present invention not be limited to the described embodiments, but that it has the full scope defined by the language of the following claims, and equivalents thereof.

Claims

WHAT IS CLAIMED IS:

1. A method of building a microfluidic device comprising: forming a main raised trace with a rectangular cross-section on a first substrate, the main raised trace having a first portion and a final portion; forming a first raised trace with a rectangular cross-section on the first substrate, wherein the first raised trace meets the final portion of the main raised trace; forming a second raised trace with a rectangular cross-section on the first substrate wherein the second raised trace meets the final portion of the main raised trace, wherein the first portion of the main raised trace is located on a first part of the first substrate, and wherein at least part of the first raised trace and at least part of the second raised trace are located on a second part of the first substrate; heating the second part of the first substrate until the cross-section of at least part of the first raised trace and the cross-section of at least part of the second raised trace become rounded, while ensuring that the first portion of the main trace is not heated to a point where its cross section becomes rounded; subsequent to the heating, pouring an uncured polymer material on top of the first substrate, the main raised trace, the first raised trace, and the second raised trace; and subsequent to the pouring, curing the polymer material to form a transparent flow layer.

2. The method of claim 1, wherein the main raised trace, the first raised trace, and the second raised trace are made from a cured photoresist material.

3. The method of claim 1, wherein the main raised trace, the first raised trace, and the second raised trace are made from cured AZ 40XT-11D photoresist material, and wherein the first substrate comprises a silicon wafer.

4. The method of claim 1, wherein the first portion of the main raised trace is located at least 1 cm away from the first raised trace and the second raised trace.

5. The method of claim 1, further comprising: aligning the flow layer to a control layer so that a first control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the first raised trace, and so that a second control surface on the control layer aligns with a part of the flow layer that corresponds to a rounded part of the second raised trace; and bonding the flow layer to the control layer subsequent to the aligning.

6. The method of claim 5, further comprising bonding the control layer to a transparent slide.

7. The method of claim 5, further comprising: prior to the aligning, forming the control layer by (a) forming a plurality of raised traces with rectangular cross-sections on a second substrate, (b) pouring an uncured elastomer material on top of the second substrate and the plurality of raised traces and subsequently curing the elastomer to form the control layer.

8. A microfluidic device comprising: a flow layer having a front surface and a rear surface wherein the rear surface of the flow layer has a main indentation that is shaped and dimensioned to form a main fluid-flow channel having an inlet, a first portion having a flat front surface, and a final portion that is downstream from the first portion, wherein the main fluid- flow channel runs along the rear surface of the flow layer, wherein the rear surface of the flow layer has a first indentation that is shaped and dimensioned to form a first fluid-flow channel that runs along the rear surface of the flow layer between (a) the final portion of the main fluid-flow channel and (b) a first outlet, and wherein at least part of the first fluid-flow channel has a rounded front surface; and a control layer bonded against the rear surface of the flow layer, wherein the control layer has a first control surface that is aligned with the part of the first fluid-flow channel that has a rounded front surface, and wherein the control layer has a first control channel aligned with the first control surface such that pressurizing the first control channel above a first pressure threshold causes the first control surface to bulge into the part of the first fluidflow channel that has a rounded front surface, which prevents fluid from flowing through the first fluid- flow channel.

9. The microfluidic device of claim 8, wherein the first portion and the final portion of the main fluid-flow channel are both straight, and wherein the first portion of the main fluidflow channel empties directly into the final portion of the main fluid-flow channel.

10. The microfluidic device of claim 8, wherein the main indentation is shaped so that the main fluid-flow channel has a serpentine second portion disposed between the first portion and the final portion.

11. The microfluidic device of claim 8, wherein the final portion of the main fluid-flow channel has a front surface that transitions smoothly from flat to rounded.

12. The microfluidic device of claim 8, wherein the entire flow layer is formed from a single piece of an elastomer material.

13. The microfluidic device of claim 8, further comprising a flat substrate bonded to a rear surface of the control layer.

14. The microfluidic device of claim 8, wherein the rear surface of the flow layer has a second indentation that is shaped and dimensioned to form a second fluid- flow channel that runs along the rear surface of the flow layer between (a) the final portion of the main fluid- flow channel and (b) a second outlet, and wherein at least a part of the second fluid-flow channel has a rounded front surface, wherein the control layer has a second control surface that is aligned with the part of the second fluid- flow channel that has a rounded front surface, and wherein the control layer has a second control channel aligned with the second control surface such that pressurizing the second control channel above a second pressure threshold causes the second control surface to bulge into the part of the second fluidflow channel that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel.

15. The microfluidic device of claim 14, wherein the entire flow layer is formed from a single piece of an elastomer material, and wherein the control layer comprises an elastomer material.

16. The microfluidic device of claim 14, wherein the entire flow layer is formed from a single piece of PDMS, and wherein the control layer comprises PDMS.

17. The microfluidic device of claim 14, further comprising a flat substrate bonded to a rear surface of the control layer.

18. A microfluidic device comprising: a substrate having a front surface, a layer of elastomer material having a front surface and a rear surface, wherein the rear surface of the layer of elastomer material is bonded to the front surface of the substrate; and a second layer of material having a front surface and a rear surface, wherein the rear surface of the second layer of material is bonded to the front surface of the layer of elastomer material, wherein the rear surface of the second layer of material has a main indentation that is shaped and dimensioned to form a main fluid-flow channel having an inlet, a first portion having a flat front surface, and a final portion that is downstream from the first portion, wherein the main fluid-flow channel runs along the rear surface of the second layer of material, wherein the rear surface of the second layer of material has a first indentation that is shaped and dimensioned to form a first fluid- flow channel that runs along the rear surface of the second layer of material between (a) the final portion of the main fluid-flow channel and (b) a first outlet, and wherein at least part of the first fluid-flow channel has a rounded front surface, wherein the rear surface of the second layer of material has a second indentation that is shaped and dimensioned to form a second fluid-flow channel that runs along the rear surface of the second layer of material between (a) the final portion of the main fluidflow channel and (b) a second outlet, and wherein at least a part of the second fluidflow channel has a rounded front surface, wherein the rear surface of the layer of elastomer material has a first indentation that forms a first control channel, wherein the first control channel is shaped, dimensioned, and aligned with the first fluid-flow channel so that pressurizing the first control channel above a first pressure threshold causes a first part of the front surface of the layer of elastomer material to bulge forward into the part of the first fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the first fluidflow channel, and wherein the rear surface of the layer of elastomer material has a second indentation that forms a second control channel, wherein the second control channel is shaped, dimensioned, and aligned with the second fluid- flow channel so that pressurizing the second control channel above a second pressure threshold causes a second part of the front surface of the layer of elastomer material to bulge forward into the part of the second fluid-flow channel that has a rounded front surface, which prevents fluid from flowing through the second fluid-flow channel.

19. The microfluidic device of claim 18, wherein the first portion and the final portion of the main fluid-flow channel are both straight, and wherein the first portion of the main fluidflow channel empties directly into the final portion of the main fluid-flow channel.

20. The microfluidic device of claim 18, wherein the main indentation is shaped so that the main fluid-flow channel has a serpentine second portion disposed between the first portion and the final portion.

21. The microfluidic device of claim 18, wherein the final portion of the main fluid- flow channel has a front surface that transitions smoothly from flat to rounded.

22. The microfluidic device of claim 18, wherein the second layer of material comprises an elastomer material.

23. The microfluidic device of claim 18, wherein the layer of elastomer material comprises PDMS, and the second layer of material comprises PDMS.

24. A microfluidic device comprising: a flow layer having a front surface and a rear surface; and a control layer bonded against the rear surface of the flow layer, wherein the flow layer has a plurality of fluid-flow channels, wherein each of the fluid-flow channels has a respective roof, wherein regions of the fluid-flow channels that are used to implement imaging have rectangular roofs, and wherein regions of the fluid-flow channels that are used to implement valve functions have curved roofs.