Generalizable transfer learning of automated tumor segmentation from cervical cancers toward a universal model for uterine malignancies in diffusion-weighted MRI

Yu-Chun Lin; Yenpo Lin; Yen-Ling Huang; Chih-Yi Ho; Hsin-Ju Chiang; Hsin-Ying Lu; Chun-Chieh Wang; Jiun-Jie Wang; Shu-Hang Ng; Chyong-Huey Lai; Gigin Lin

doi:10.1186/s13244-022-01356-8

Generalizable transfer learning of automated tumor segmentation from cervical cancers toward a universal model for uterine malignancies in diffusion-weighted MRI

Insights Imaging. 2023 Jan 24;14(1):14. doi: 10.1186/s13244-022-01356-8.

Authors

Yu-Chun Lin^{1

2

3}, Yenpo Lin¹, Yen-Ling Huang¹, Chih-Yi Ho¹, Hsin-Ju Chiang^{1

3}, Hsin-Ying Lu^{1

3}, Chun-Chieh Wang^{2

4}, Jiun-Jie Wang^{1

2}, Shu-Hang Ng¹, Chyong-Huey Lai⁵, Gigin Lin^{6

7

8}

Affiliations

¹ Department of Medical Imaging and Intervention, Chang Gung Memorial Hospital at Linkou and Keelung, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan.
² Department of Medical Imaging and Radiological Sciences, Chang Gung University, Taoyuan, 33302, Taiwan.
³ Clinical Metabolomics Core Laboratory, Chang Gung Memorial Hospital at Linkou, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan.
⁴ Department of Radiation Oncology, Chang Gung Memorial Hospital at Linkou and Chang Gung University, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan.
⁵ Gynecologic Cancer Research Center, Department of Obstetrics and Gynecology, Chang Gung Memorial Hospital at Linkou and Chang Gung University, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan.
⁶ Department of Medical Imaging and Intervention, Chang Gung Memorial Hospital at Linkou and Keelung, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan. giginlin@cgmh.org.tw.
⁷ Clinical Metabolomics Core Laboratory, Chang Gung Memorial Hospital at Linkou, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan. giginlin@cgmh.org.tw.
⁸ Gynecologic Cancer Research Center, Department of Obstetrics and Gynecology, Chang Gung Memorial Hospital at Linkou and Chang Gung University, 5 Fuhsing St., Guishan, Taoyuan, 33382, Taiwan. giginlin@cgmh.org.tw.

Abstract

Purpose: To investigate the generalizability of transfer learning (TL) of automated tumor segmentation from cervical cancers toward a universal model for cervical and uterine malignancies in diffusion-weighted magnetic resonance imaging (DWI).

Methods: In this retrospective multicenter study, we analyzed pelvic DWI data from 169 and 320 patients with cervical and uterine malignancies and divided them into the training (144 and 256) and testing (25 and 64) datasets, respectively. A pretrained model was established using DeepLab V3 + from the cervical cancer dataset, followed by TL experiments adjusting the training data sizes and fine-tuning layers. The model performance was evaluated using the dice similarity coefficient (DSC).

Results: In predicting tumor segmentation for all cervical and uterine malignancies, TL models improved the DSCs from the pretrained cervical model (DSC 0.43) when adding 5, 13, 26, and 51 uterine cases for training (DSC improved from 0.57, 0.62, 0.68, 0.70, p < 0.001). Following the crossover at adding 128 cases (DSC 0.71), the model trained by combining data from adding all the 256 patients exhibited the highest DSCs for the combined cervical and uterine datasets (DSC 0.81) and cervical only dataset (DSC 0.91).

Conclusions: TL may improve the generalizability of automated tumor segmentation of DWI from a specific cancer type toward multiple types of uterine malignancies especially in limited case numbers.

Keywords: Deep learning; Magnetic resonance imaging; Uterine neoplasms.

Abstract

Grants and funding