A major problem in cell-labeling via photoconversion lies in the time-consuming process of manually searching and tracing target cells, leading to delay of research results that can benefit modern medicine.
Our project wants to achieve the automation of phenotypic cell sorting, cell contouring, and photoconversion in a 3D-cell culture. Nowadays, 3D-cell cultures are becoming the new standard in cell culture due to the better representation of the in vivo environment. On the other hand, the tools for 3D-cell culture are still in development. Researchers are still manually sorting, outlining, and photo-converting the cells, which takes hours or even days to finish. We are taking steps to automate this process so that researchers can spend their time on other tasks. Also, we hope to identify new phenotypes and understand the underlying mechanisms of cancer migration and progression.
Our software needs to automate the cell segmentation process for researchers. Therefore, having an assessment of its execution speed is important. To evaluate the execution speed of the software, its execution speed will be compared to the time it takes to perform the process manually. Success will be determined if the execution speed is significantly less than the time it takes to manually perform the process. If the execution speed was not reduced, the time that the researcher spends to do the same task can be compared. Execution speed must be fast in order to expedite and throughput research results. Otherwise, the researcher should be able to spend their time on other tasks instead of manually proceeding this process.
Identifying the phenotype of the cancer cells in a 3D collage cell culture should be reliable and accurate. The system must be able to discriminate between the cancer cells and any contaminants that may exist in the images, allowing the automated process to produce accurate research results. To evaluate the reliability of the software, its error rate must be determined by counting the number of times it accurately photoconverted ideal cells and the number of times it photoconverted off-target cells. The software must also be able to accurately photoconvert target cancer cell populations that will be molecularly analyzed to improve cancer therapeutics.
The main goal is that the data obtained from the photoconversion process must be understandable by a researcher. In general, the system must provide information to the researcher regarding which specific areas it will be photoconverting with the laser as well as information (size, circularity, etc.) about other cells in the 3D z-stacks, allowing researchers to potentially come back to the data to re-analyze it. To evaluate interpretability, multiple end-users will be asked to determine if the output of the software can be understood with minimal instruction. The system must also be able to record which exact tumors it has identified so that the data can potentially be re-analyzed.