MRI-based automatic segmentation of rectal cancer using 2D U-Net on two independent cohorts.

Autor: Knuth F; Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway., Adde IA; Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway., Huynh BN; Faculty of Science and Technology, Norwegian University of Life Sciences, Ås, Norway., Groendahl AR; Faculty of Science and Technology, Norwegian University of Life Sciences, Ås, Norway., Winter RM; Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway., Negård A; Department of Radiology, Akershus University Hospital, Lørenskog, Norway.; Institute of Clinical Medicine, University of Oslo, Oslo, Norway., Holmedal SH; Department of Radiology, Akershus University Hospital, Lørenskog, Norway., Meltzer S; Department of Oncology, Akershus University Hospital, Lørenskog, Norway., Ree AH; Institute of Clinical Medicine, University of Oslo, Oslo, Norway.; Department of Oncology, Akershus University Hospital, Lørenskog, Norway., Flatmark K; Institute of Clinical Medicine, University of Oslo, Oslo, Norway.; Department of Gastroenterological Surgery, Oslo University Hospital, Oslo, Norway., Dueland S; Department of Oncology, Oslo University Hospital, Oslo, Norway., Hole KH; Institute of Clinical Medicine, University of Oslo, Oslo, Norway.; Division of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway., Seierstad T; Division of Radiology and Nuclear Medicine, Oslo University Hospital, Oslo, Norway., Redalen KR; Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway., Futsaether CM; Faculty of Science and Technology, Norwegian University of Life Sciences, Ås, Norway.
Jazyk: angličtina
Zdroj: Acta oncologica (Stockholm, Sweden) [Acta Oncol] 2022 Feb; Vol. 61 (2), pp. 255-263. Date of Electronic Publication: 2021 Dec 17.
DOI: 10.1080/0284186X.2021.2013530
Abstrakt: Background: Tumor delineation is time- and labor-intensive and prone to inter- and intraobserver variations. Magnetic resonance imaging (MRI) provides good soft tissue contrast, and functional MRI captures tissue properties that may be valuable for tumor delineation. We explored MRI-based automatic segmentation of rectal cancer using a deep learning (DL) approach. We first investigated potential improvements when including both anatomical T2-weighted (T2w) MRI and diffusion-weighted MR images (DWI). Secondly, we investigated generalizability by including a second, independent cohort.
Material and Methods: Two cohorts of rectal cancer patients (C1 and C2) from different hospitals with 109 and 83 patients, respectively, were subject to 1.5 T MRI at baseline. T2w images were acquired for both cohorts and DWI (b-value of 500 s/mm 2 ) for patients in C1. Tumors were manually delineated by three radiologists (two in C1, one in C2). A 2D U-Net was trained on T2w and T2w + DWI. Optimal parameters for image pre-processing and training were identified on C1 using five-fold cross-validation and patient Dice similarity coefficient (DSC p ) as performance measure. The optimized models were evaluated on a C1 hold-out test set and the generalizability was investigated using C2.
Results: For cohort C1, the T2w model resulted in a median DSC p of 0.77 on the test set. Inclusion of DWI did not further improve the performance (DSC p 0.76). The T2w-based model trained on C1 and applied to C2 achieved a DSC p of 0.59.
Conclusion: T2w MR-based DL models demonstrated high performance for automatic tumor segmentation, at the same level as published data on interobserver variation. DWI did not improve results further. Using DL models on unseen cohorts requires caution, and one cannot expect the same performance.
Databáze: MEDLINE
Externí odkaz:
Nepřihlášeným uživatelům se plný text nezobrazuje