Hua Ye

AI-assisted cervical cytology precancerous screening for high-risk population in resource-limited regions using a compact microscope

Insufficient coverage of cervical cytology screening in resource-limited areas remains a major bottleneck for women's health, as traditional centralized methods require significant investment and many qualified pathologists. Using consumer-grade electronic hardware and aspherical lenses, we design an ultra-low-cost and compact microscope. Given the microscope's low resolution, which hinders accurate identification of lesion cells in cervical samples, we train a coarse instance classifier to screen and extract feature sequences of the top 200 instances containing potential lesions from a slide. We further develop Att-Transformer to focus on and integrate the sparse lesion information from these sequences, enabling slide grading. Our model is trained and validated using 3510 low-resolution slides from female patients at four hospitals, and subsequently evaluated on four independent datasets. The system achieves area under the receiver operating characteristic curve values of 0.87 and 0.89 for detecting squamous intraepithelial lesions on 364 slides from female patients at two external primary hospitals, 0.89 on 391 newly collected slides from female patients at the original four hospitals, and 0.85 on 570 human papillomavirus positive slides from female patients. These findings demonstrate the feasibility of our AI-assisted approach for effective detection of high-risk cervical precancer among women in resource-limited regions.

Serum Autoantibodies against LRDD, STC1, and FOXA1 as Biomarkers in the Detection of Ovarian Cancer

Purpose. This study is aimed at evaluating serum autoantibodies against four tumor-associated antigens, including LRDD, STC1, FOXA1, and EDNRB, as biomarkers in the immunodiagnosis of ovarian cancer (OC). Methods. The autoantibodies against LRDD, STC1, FOXA1, and EDNRB were measured using an enzyme-linked immunosorbent assay (ELISA) in 94 OC patients and 94 normal healthy controls (NHC) in the research group. In addition, the diagnostic values of different autoantibodies were validated in another independent validation group, which comprised 136 OC patients, 136 NHC, and 181 patients with benign ovarian diseases (BOD). Results. In the research group, autoantibodies against LRDD, STC1, and FOXA1 had higher serum titer in OC patients than NHC ( P < 0.001 ). The area under receiver operating characteristic curves (AUCs) of these three autoantibodies were 0.910, 0.879, and 0.817, respectively. In the validation group, they showed AUCs of 0.759, 0.762, and 0.817 and sensitivities of 49.3%, 42.7%, and 48.5%, respectively, at specificity over 90% for discriminating OC patients from NHC. For discriminating OC patients from BOD, they showed AUCs of 0.718, 0.729, and 0.814 and sensitivities of 47.1%, 39.0%, and 51.5%, respectively, at specificity over 90%. The parallel analyses demonstrated that the combination of anti-LRDD and anti-FOXA1 autoantibodies achieved the optimal diagnostic performance with the sensitivity of 58.1% at 87.5% specificity and accuracy of 72.8%. The positive rate of the optimal autoantibody panel improved from 62.4% to 87.1% when combined with CA125 in detecting OC patients. Conclusion. Serum autoantibodies against LRDD, STC1, and FOXA1 have potential diagnostic values in detecting OC.