Artificial intelligence-assisted triage of cervical biopsies using the low-cost, 3D-printed OpenFlexure Microscope: a paired comparative feasibility study.

OBJECTIVE

Artificial intelligence has been used in pathology to analyze whole-slide images for the triage of biopsies for pathologist review. However, this has been performed with prohibitively costly slide scanners that may be unavailable in low- and middle-income countries, where the number of trained pathologists is limited. We evaluated whether a 3D-printed slide scanner (OpenFlexure Microscope) could create images of sufficient quality that could be used for artificial intelligence triage of cervical biopsies and compared them to images created by a commercial slide scanner.

METHODS

Cervical biopsy slides from 275 specimens at the University of Michigan were scanned using an OpenFlexure Microscope and Aperio GT450; 186 were used for the model training/validation set and the remaining 89 for the holdout set. Post-capture edits, including pixel-resolution standardization and color correction, were performed to ensure minimal differences between the 2 images. Our team created and tested 5 models using open-source software to classify biopsies into high- and low-risk groups for high-grade squamous intra-epithelial neoplasia.

RESULTS

The holdout set comprised 29 high- and 60 low-risk cases. Metrics ranged across models: area under the receiver operating characteristic curve 0.71 to 0.80, accuracy 0.64 to 0.72, precision 0.45 to 0.55, F1 score 0.45 to 0.64, and sensitivity 0.45 to 0.76. The 95% confidence intervals for most metrics overlapped across all 5 models, indicating a lack of statistical power to support any one model outperforming the other models.

CONCLUSIONS

This comparative feasibility study suggests that an open-source 3D-printed slide-scanning device can produce images of sufficient quality for artificial intelligence-assisted cervical biopsy triage to aid pathologists in flagging high-risk cases that require expedited reviews. Additional research with larger multi-site cohorts from low- and middle-income countries, where tissue processing and disease prevalence differ, is needed to assess real-world applications.