Automatic Recognition of Abdominal Organs in Ultrasound Images based on Deep Neural Networks and K-Nearest-Neighbor Classification

TL;DR

A classification method that combines the deep learning techniques and k-Nearest-Neighbor (k-NN) classification to automatically recognize various abdominal organs in the ultra-sound images in real time with an accuracy of 96.67%.

Abstract

Abdominal ultrasound imaging has been widely used to assist in the diagnosis and treatment of various abdominal organs. In order to shorten the examination time and reduce the cognitive burden on the sonographers, we present a classification method that combines the deep learning techniques and k-Nearest-Neighbor (k-NN) classification to automatically recognize various abdominal organs in the ultrasound images in real time. Fine-tuned deep neural networks are used in combination with PCA dimension reduction to extract high-level features from raw ultrasound images, and a k-NN classifier is employed to predict the abdominal organ in the image. We demonstrate the effectiveness of our method in the task of ultrasound image classification to automatically recognize six abdominal organs. A comprehensive comparison of different configurations is conducted to study the influence of different feature extractors and classifiers on the classification accuracy. Both quantitative and qualitative results show that with minimal training effort, our method can "lazily" recognize the abdominal organs in the ultrasound images in real time with an accuracy of 96.67%. Our implementation code is publicly available at: https://github.com/LeeKeyu/abdominal_ultrasound_classification.

Figures (6)

• Figure 1: The overall workflow of the proposed method for abdominal organ recognition from ultrasound images.
• Figure 2: Abdominal ultrasound images of 6 different organs (a) bladder, (b) bowel, (c) gallbladder, (d) kidney, (e) liver, and (f) spleen.
• Figure 3: Learning curves of the ResNet-50, ResNet-101, ResNet-152, DenseNet-121, DenseNet-169 and DenseNet-201 models during fine-tuning on our data for abdominal organ classification, which indicate the (a) training loss and (b) training accuracy as functions of the training epochs.
• Figure 4: Classification performance using different combinations of features extractors (pretrained deep NN models, fine-tuned deep NN models, and fine-tuned deep NN models with PCA dimension reduction) and classifiers (k-NN-based classifiers and fine-tuned FC layer classifier).
• Figure 5: Confusion matrices of the methods (a) "fine-tuned DenseNet-121 feature extractor + FC layer classifier" and (b) "fine-tuned DenseNet-121+PCA feature extractor + k-NN classifier" for the classification of abdominal ultrasound images in the test set.