Enhancing Robustness of Deep Learning to Batch Effects from Multi-site Data for Segmentation of Clinically...

Deep learning (DL) has shown promise in segmenting clinically significant prostate cancer (csPCa) on MRI. However, batch effects arising from multi-site variations impact their generalizability. This study investigates the impact of training data harmonization, diversity, and few-shot fine-tuning on DL in segmenting csPCa on multi-site MRI. 3 T prostate MRI from N = 1822 patients of four sites (public: D₁, N = 1500; D₂, N = 157; institutional: D₃, N = 47; D₄, N = 118) was leveraged. Bi-parametric MRI (T2-weighted, ADC) was harmonized with csPCa lesions delineated by expert radiologists. nnU-Net DL models for csPCa segmentation on MRI were trained separately (C₁ on D₁; C₂ on D₂) and jointly (C₃ on D₁-D₂). The jointly trained C₃ model was then fine-tuned on D₃ and D₄ using few-shot learning in increments of 50%, 75%, and 100% of train data. Models were primarily evaluated on holdout test sets using sensitivity primarily, in addition to AUC, DSC, precision, and Hausdorff distance (paired t-tests for evaluating significance). Batch effects persisted across datasets despite harmonization. For csPCa segmentation on D₁/D₂ test sets, C₁ achieved sensitivities of 0.51 ± 0.34/0.11 ± 0.17, C₂ achieved 0.23 ± 0.31/0.17 ± 0.22, while C₃ improved to 0.48 ± 0.36/0.33 ± 0.26. On D₃/D₄ test sets, C₃ achieved zero-shot sensitivities of 0.19 ± 0.22/0.37 ± 0.35. Few-shot fine-tuning on D₃ and D₄ improved sensitivity to 0.32 ± 0.26/0.54 ± 0.30 on D₃/D₄ test sets after using 100% of available fine-tuning samples. On subset analyses, lesions > 0.5 cm³ had consistently higher segmentation performance compared to smaller lesions < 0.5 cm³. Diverse multi-site training improves robustness of DL models for csPCa segmentation. Pre-trained models perform modestly on institutional MRI datasets under zero-shot inference, but few-shot fine-tuning on target sites enhances performance.