Drazinos Petros, Gatos Ilias, Katsakiori Paraskevi F, Tsantis Stavros, Syrmas Efstratios, Spiliopoulos Stavros, Karnabatidis Dimitris, Theotokas Ioannis, Zoumpoulis Pavlos, Hazle John D, Kagadis George C
3DMI Research Group, Department of Medical Physics, School of Medicine, University of Patras, Rion, GR 26504, Greece; Diagnostic Echotomography SA, Kifissia, GR 14561, Greece.
3DMI Research Group, Department of Medical Physics, School of Medicine, University of Patras, Rion, GR 26504, Greece.
Phys Med. 2025 Jan;129:104862. doi: 10.1016/j.ejmp.2024.104862. Epub 2024 Dec 2.
BACKGROUND/INTRODUCTION: To evaluate the performance of pre-trained deep learning schemes (DLS) in hepatic steatosis (HS) grading of Non-Alcoholic Fatty Liver Disease (NAFLD) patients, using as input B-mode US images containing right kidney (RK) cortex and liver parenchyma (LP) areas indicated by an expert radiologist.
A total of 112 consecutively enrolled, biopsy-validated NAFLD patients underwent a regular abdominal B-mode US examination. For each patient, a radiologist obtained a B-mode US image containing RK cortex and LP and marked a point between the RK and LP, around which a window was automatically cropped. The cropped image dataset was augmented using up-sampling, and the augmented and non-augmented datasets were sorted by HS grade. Each dataset was split into training (70%) and testing (30%), and fed separately as input to InceptionV3, MobileNetV2, ResNet50, DenseNet201, and NASNetMobile pre-trained DLS. A receiver operating characteristic (ROC) analysis of hepatorenal index (HRI) measurements by the radiologist from the same cropped images was used for comparison with the performance of the DLS.
With the test data, the DLS reached 89.15 %-93.75 % accuracy when comparing HS grades S0-S1 vs. S2-S3 and 79.69 %-91.21 % accuracy for S0 vs. S1 vs. S2 vs. S3 with augmentation, and 80.45-82.73 % accuracy when comparing S0-S1 vs. S2-S3 and 59.54 %-63.64 % accuracy for S0 vs. S1 vs. S2 vs. S3 without augmentation. The performance of radiologists' HRI measurement after ROC analysis was 82 %, 91.56 %, and 96.19 % for thresholds of S ≥ S1, S ≥ S2, and S = S3, respectively.
All networks achieved high performance in HS assessment. DenseNet201 with the use of augmented data seems to be the most efficient supplementary tool for NAFLD diagnosis and grading.
背景/引言：为了评估预训练深度学习方案（DLS）在非酒精性脂肪性肝病（NAFLD）患者肝脂肪变性（HS）分级中的性能，使用由专业放射科医生标注的包含右肾（RK）皮质和肝实质（LP）区域的B超图像作为输入。
总共112例连续入选且经活检验证的NAFLD患者接受了常规腹部B超检查。对于每位患者，放射科医生获取一张包含RK皮质和LP的B超图像，并在RK和LP之间标记一个点，围绕该点自动裁剪出一个窗口。裁剪后的图像数据集通过上采样进行扩充，扩充后的数据集和未扩充的数据集按HS分级进行分类。每个数据集被分为训练集（70%）和测试集（30%），并分别作为输入提供给InceptionV3、MobileNetV2、ResNet50、DenseNet201和NASNetMobile预训练的DLS。对放射科医生从相同裁剪图像中测量的肝肾指数（HRI）进行受试者操作特征（ROC）分析，以与DLS的性能进行比较。
对于测试数据，在比较HS分级S0 - S1与S2 - S3时，DLS的准确率达到89.15% - 93.75%；在比较S0 vs. S1 vs. S2 vs. S3时，扩充数据后的准确率为79.69% - 91.21%，未扩充数据时，在比较S0 - S1与S2 - S3时准确率为80.45 - 82.73%，在比较S0 vs. S1 vs. S2 vs. S3时准确率为59.54% - 63.64%。ROC分析后，放射科医生测量HRI的性能在阈值S≥S1、S≥S2和S = S3时分别为82%、91.56%和96.19%。
所有网络在HS评估中均表现出高性能。使用扩充数据的DenseNet201似乎是NAFLD诊断和分级最有效的辅助工具。
