Hepatocellular carcinoma (HCC) constitutes a prominent global health challenge. According to the American Association for the Study of Liver Diseases (AASLD) guideline HCC can be diagnosed by imaging examination. However, it shows that contrast-enhanced CT has limited accuracy in the diagnosis of HCC, particularly, small-size HCC lesions (≤ 20 mm) are the most difficult to identify with CT demonstrating sensitivity = 64% [1]. An Artificial Intelligence (AI) algorithm that can analyze liver CT images and localize HCC lesions would be valuable for patients at risk...

MethodsThe proposed DL model combines a specifically designed convolutional 3D classifier and a state-of-the-art 3D detector (3D-RetinaUnet). It leverages the arterial and portal venous phases of CT scans as input and provides the locations of HCC lesions as output.Our global workflow (Figure 1) has two main branches, the one with Median in-house developed Multiscale Autoencoders [5,6] and the one with the Retina 3D-Unet branch. Each branch processes the arterial and portal phases together and proposes a set of 3D bounding boxes as candidates for the...

Figure 7 displays the FROC curves for 3D Intersection over Union (IoU). The IoU is shown in Figure 6 and measures how well the detection was localised. A correct detection was recognized when the IoU between the detection box and the ground truth box was higher than a certain threshold (0.1, 0.2 and 0.3 in our case). IoU=1.0 means a perfect match between the detection and the radiologists’ annotation.[Fig 6]At 3.0 false positive per scan (FP/s) we have sensitivity of 0.945, 0.928, 0.900, respectively for...

Our DL model could potentially increase HCC diagnosis accuracy compared to the literature reported radiologist performances (Sensitivity of 64% for diagnosing small HCCs [1]). The next step is to further refine our DL model and increase the training dataset with more at-risk patients with benign and malignant liver nodules and masses. Our goal is to develop an end-to-end Computer-Aided Detection and Computer-Aided Diagnosis (CADe/CADx) that will improve the diagnostic accuracy for HCC, particularly for localising small HCCs on CT images. In this paper, we achieved...

O. Lucidarme: Other: Median Technologies V. Paradis: Nothing to disclose C. Guettier: Nothing to disclose I. Brocheriou: Nothing to disclose J. Shen: Employee: Median Technologies S. Poullot: Employee: Median Technologies V. K. LE: Employee: Median Technologies V. Vilgrain: Nothing to disclose M. Lewin-Zeitoun: Nothing to disclose

[1] Roberts, Lewis R., et al. "Imaging for the diagnosis of hepatocellular carcinoma: a systematic review and meta‐analysis." Hepatology 67.1 (2018): 401-421.[2] Wang M, Fu F, Zheng B, et al. “Development of an AI system for accurately diagnose hepatocellular carcinoma from computed tomography imaging data.” Translational Therapeutics British Journal of Cancer 125 (2021):1111–1121.[3] Kim DW, Lee G, Kim SY, et al. “Deep learning-based algorithm to detect primary hepatic malignancy in multiphase CT of patients at high risk for HCC.” Eur Radiol 31 (2021):7047–7057.[4] LUCIDARME, O.,...