3-D RoI-Aware U-Net for Accurate and Efficient Colorectal Tumor Segmentation

Abstract

Segmentation of colorectal cancerous regions from 3DMagnetic Resonance (MR) images is a crucial procedure for radiotherapy which conventionally requires accurate delineation of tumour boundaries at an expense of labor, time and reproducibility. While deep learning based methods serve good baselines in 3D image segmentation tasks, small applicable patch size limits effective receptive field and degrades segmentation performance. In addition, Regions of interest (RoIs) localization from large whole volume 3D images serves as a preceding operation that brings about multiple benefits in terms of speed, target completeness, reduction of false positives. Distinct from sliding window or non-joint localizationsegmentation based models, we propose a novel multitask framework referred to as 3D RoI-aware U-Net (3D RU-Net), for RoI localization and in-region segmentation where the two tasks share one backbone encoder network. With the region proposals from the encoder, we crop multi-level RoI in-region features from the encoder to form a GPU memory-efficient decoder for detailpreserving segmentation and therefore enlarged applicable volume size and effective receptive field. To effectively train the model, we designed a Dice formulated loss function for the global-to-local multi-task learning procedure. Based on the efficiency gains demonstrated by the proposed method, we went on to ensemble models with different receptive fields to achieve even higher performance costing minor extra computational expensiveness. Extensive experiments were subsequently conducted on 64 cancerous cases with a four-fold cross-validation, and the results showed significant superiority in terms of accuracy and efficiency over conventional state-of-the art frameworks. In conclusion, the proposed method has a huge potential for extension to other 3D object segmentation tasks from medical images due to its inherent generalizability. The code for the proposed method is publicly available.3D CNN, region of interest, multi-task learning, tumor segmentation, colorectal cancer.

Embracing imperfect datasets: A review of deep learning solutions for medical image segmentation

“…Similarly, Jaeger et al (2018); Huang et al (2018) propose a joint segmentation and detection framework. Huang et al (2018) use their method for colorectal tumor segmentation in MRI volumes.…”

“…Similarly, Jaeger et al (2018); Huang et al (2018) propose a joint segmentation and detection framework. Huang et al (2018) use their method for colorectal tumor segmentation in MRI volumes. Their proposed model resembles Mask- RCNN He et al (2017), in which a global image encoder network detects regions of interest (ROIs), and a local decoder performs tumor segmentation on the proposed regions.…”

Toward human intervention-free clinical diagnosis of intracranial aneurysm via deep neural network

“…This can be ignored in some applications like tumor detection in pathology slices but is very important for the lesions that have a global position tendency in the body, like IAs. Although there are some works considering that multi-level feature fusion can achieve a plausible effect on targets with different sizes, 27 , 28 they still adopt the whole image as input, which means they cannot accommodate such huge CTA images in one glance. Some medical imaging models do consider the location information, 29 , 30 , 31 but they just directly use the coordinate values without delicately extracting and representing the location information, thus they are closely attached to the data they use.…”

MMFNet: A multi-modality MRI fusion network for segmentation of nasopharyngeal carcinoma

“…The preprocessing mainly contains intensity normalization and ROIs cropping. We utilize the intra-body intensity normalization proposed in [43], which effectively deals with the differences caused by imaging configurations and the influences of inconsistent body-to-background ratios. After normalization, according to the distribution histogram of normalized data, we Figure 6.…”