Splenomegaly Segmentation on Multi-Modal MRI Using Deep Convolutional Networks

Abstract

The findings of splenomegaly, abnormal enlargement of the spleen, is a non-invasive clinical biomarker for liver and spleen disease. Automated segmentation methods are essential to efficiently quantify splenomegaly from clinically acquired abdominal magnetic resonance imaging (MRI) scans. However, the task is challenging due to (1) large anatomical and spatial variations of splenomegaly, (2) large inter-and intra-scan intensity variations on multi-modal MRI, and (3) limited numbers of labeled splenomegaly scans. In this paper, we propose the Splenomegaly Segmentation Network (SS-Net) to introduce the deep convolutional neural network (DCNN) approaches in multi-modal MRI splenomegaly segmentation. Large convolutional kernel layers were used to address the spatial and anatomical variations, while the conditional generative adversarial networks (GAN) were employed to leverage the segmentation performance of SS-Net in an end-to-end manner. A clinically acquired cohort containing both T1-weighted (T1w) and T2weighted (T2w) MRI splenomegaly scans was used to train and evaluate the performance of multiatlas segmentation (MAS), 2D DCNN networks, and a 3D DCNN network. From the experimental results, the DCNN methods achieved superior performance to the state-of-the-art MAS method. The proposed SS-Net method achieved the highest median and mean Dice scores among investigated baseline DCNN methods.

Outlier guided optimization of abdominal segmentation

“…2, accordingly). The other 120 subjects were deidentified splenomegaly patients with manually labeled spleens [7], all of which were used for the training set (Group 2). A separate clinical dataset of 6,317 deidentified clinical scans were used for assessing the baseline algorithm.…”

Multiscale matters for part segmentation of instruments in robotic surgery

“…Encoders with limited receptive fields or small kernels may not be suitable for representing semantic pixels. Therefore, convolution operations in a fully convolutional neural network are replaced with dilated convolutions [7] or deformable ones [41], and large kernels in a global convolutional network simultaneously localise and classify pixels [29, 42, 43]. However, encoders with small kernels might be faster for dense pixel classification.…”

“…Additionally, structure priors in images can be efficiently fused using multi‐modal inputs. For example, spatial ranking maps are supervised by panoptic segmentation labels to alleviate overlapping problems among various classes [42], high‐resolution feature maps are refined with inputs from multiple paths [49], streams, respectively, given stacked optical flows and colour images model motions and appreances [50], and gated convolutional layers enforce boundary information merely processed in a shape stream [32]; images in axial, coronal, and sagittal views are independently segmented and fused into a single result with union operations [43].…”

“…Inspired by generative adversarial networks, generated masks are usually judged with a discriminator [62, 63]. For example, ground truth masks and predicted ones are separated with a classifier in image level [6], patch level [43], or both [60]. Moreover, their consistencies can be supervised with earth‐mover distances in the Wasserstein generative adversarial networks [64, 65].…”

REF-Net: Robust, Efficient, and Fast Network for Semantic Segmentation Applications Using Devices With Limited Computational Resources

“…tion fields, namely, medicine [2]- [4], biomedicine [5]- [7], geo-sensing [8]- [10], fashion [11]- [13], and autonomous driving [14]- [16].…”