Evaluation of ultrasonic fibrosis diagnostic system using convolutional network for ordinal regression

Artificial Intelligence and Image Analysis-Assisted Diagnosis for Fibrosis Stage of Metabolic Dysfunction-Associated Steatotic Liver Disease Using Ultrasonography: A Pilot Study

BACKGROUND/OBJECTIVES

Elastography increased the diagnostic accuracy of liver fibrosis. However, several challenges persist, including the widespread utilization of equipment, difficulties in measuring certain cases, and the influence of viscosity factors. A rough surface and a blunted hepatic margin have long been acknowledged as valuable characteristics indicative of hepatic fibrosis. The objective of this study was to conduct an image analysis and quantitative assessment of the contour of the sagittal section of the left lobe of the liver.

METHODS

Between February and October 2020, 486 consecutive outpatients underwent ultrasound examinations at our hospital. A total of 214 images were manually annotated by delineating the liver contour to create annotation images. U-Net was employed for liver segmentation, with the dataset divided into training (n = 128), testing (n = 42), and validation (n = 44) subsets. Additionally, 43 Metabolic Dysfunction Associated Steatotic Liver Disease (MASLD) cases with pathology data from between 2015 and 2020 were included. Segmentation was performed using the program developed in the first step. Subsequently, shape analysis was conducted using ImageJ.

RESULTS

Liver segmentation exhibited high accuracy, as indicated by Dice loss of 0.044, Intersection over Union of 0.935, and an F score of 0.966. The accuracy of the classification of the liver surface as smooth or rough via ResNet 50 was 84.6%. Image analysis showed MinFeret and Minor correlated with liver fibrosis stage (p = 0.046, 0.036, respectively). Sensitivity, specificity, and AUROC of Minor for ≥F3 were 0.571, 0.862, and 0.722, respectively, and F4 were 1, 0.600, and 0.825, respectively.

CONCLUSION

Deep learning segmentation of the sagittal cross-sectional contour of the left lobe of the liver demonstrated commendable accuracy. The roughness of the liver surface was correctly judged by artificial intelligence. Image analysis showed the thickness of the left lobe inversely correlated with liver fibrosis stage.

Convolutional neural network classification of ultrasound images by liver fibrosis stages based on echo-envelope statistics

Introduction: Assessing the stage of liver fibrosis during the diagnosis and follow-up of patients with diffuse liver disease is crucial. The tissue structure in the fibrotic liver is reflected in the texture and contrast of an ultrasound image, with the pixel brightness indicating the intensity of the echo envelope. Therefore, the progression of liver fibrosis can be evaluated non-invasively by analyzing ultrasound images.

Methods: A convolutional-neural-network (CNN) classification of ultrasound images was applied to estimate liver fibrosis. In this study, the colorization of the ultrasound images using echo-envelope statistics that correspond to the features of the images is proposed to improve the accuracy of CNN classification. In the proposed method, the ultrasound image is modulated by the 3rd- and 4th-order moments of pixel brightness. The two modulated images and the original image were then synthesized into a color image of RGB representation.

Results and Discussion: The colorized ultrasound images were classified via transfer learning of VGG-16 to evaluate the effect of colorization. Of the 80 ultrasound images with liver fibrosis stages F1–F4, 38 images were accurately classified by the CNN using the original ultrasound images, whereas 47 images were classified by the proposed method.

A Hierarchical Siamese Network for Noninvasive Staging of Liver Fibrosis Based on US Image Pairs of the Liver and Spleen

Due to the heterogeneity of ultrasound (US) images and the indeterminate US texture of liver fibrosis (LF), automatic evaluation of LF based on US images is still challenging. Thus, this study aimed to propose a hierarchical Siamese network that combines the information from liver and spleen US images to improve the accuracy of LF grading. There were two stages in the proposed method. In stage one, a dual-channel Siamese network was trained to extract features from paired liver and spleen patches that were cropped from US images to avoid vascular interferences. Subsequently, the L1 distance was used to quantify the liver-spleen differences (LSDs). In stage two, the pretrained weights from stage one were transferred into the Siamese feature extractor of the LF staging model, and a classifier was trained using the fusion of the liver and LSD features for LF staging. This study was retrospectively conducted on US images of 286 patients with histologically proven liver fibrosis stages. Our method achieved a precision and sensitivity of 93.92% and 91.65%, respectively, for cirrhosis (S4) diagnosis, which is about 8% higher than that of the baseline model. The accuracy of the advanced fibrosis (≥S3) diagnosis and the multi-staging of fibrosis (≤S2 vs. S3 vs. S4) both improved about 5% to reach 90.40% and 83.93%, respectively. This study proposed a novel method that combined hepatic and splenic US images and improved the accuracy of LF staging, which indicates the great potential of liver-spleen texture comparison in noninvasive assessment of LF based on US images.

An ensemble learning method based on ordinal regression for COVID-19 diagnosis from chest CT

Coronavirus disease 2019 (COVID-19) has brought huge losses to the world, and it remains a great threat to public health. X-ray computed tomography (CT) plays a central role in the management of COVID-19. Traditional diagnosis with pulmonary CT images is time-consuming and error-prone, which could not meet the need for precise and rapid COVID-19 screening. Nowadays, deep learning (DL) has been successfully applied to CT image analysis, which assists radiologists in workflow scheduling and treatment planning for patients with COVID-19. Traditional methods use cross-entropy as the loss function with a Softmax classifier following a fully-connected layer. Most DL-based classification methods target intraclass relationships in a certain class (early, progressive, severe, or dissipative phases), ignoring the natural order of different phases of the disease progression,i.e.,from an early stage and progress to a late stage. To learn both intraclass and interclass relationships among different stages and improve the accuracy of classification, this paper proposes an ensemble learning method based on ordinal regression, which leverages the ordinal information on COVID-19 phases. The proposed method uses multi-binary, neuron stick-breaking (NSB), and soft labels (SL) techniques, and ensembles the ordinal outputs through a median selection. To evaluate our method, we collected 172 confirmed cases. In a 2-fold cross-validation experiment, the accuracy is increased by 22% compared with traditional methods when we use modified ResNet-18 as the backbone. And precision, recall, andF1-score are also improved. The experimental results show that our proposed method achieves a better classification performance than the traditional methods, which helps establish guidelines for the classification of COVID-19 chest CT images.