Focal Liver Lesion MRI Feature Identification Using Efficientnet and MONAI: A Feasibility Study.

Liver tumors constitute a major part of the global disease burden, often making regular imaging follow-up necessary. Recently, deep learning (DL) has increasingly been applied in this research area. How these methods could facilitate report writing is still a question, which our study aims to address by assessing multiple DL methods using the Medical Open Network for Artificial Intelligence (MONAI) framework, which may provide clinicians with preliminary information about a given liver lesion. For this purpose, we collected 2274 three-dimensional images of lesions, which we cropped from gadoxetate disodium enhanced T1w, native T1w, and T2w magnetic resonance imaging (MRI) scans. After we performed training and validation using 202 and 65 lesions, we selected the best performing model to predict features of lesions from our in-house test dataset containing 112 lesions. The model (EfficientNetB0) predicted 10 features in the test set with an average area under the receiver operating characteristic curve (standard deviation), sensitivity, specificity, negative predictive value, positive predictive value of 0.84 (0.1), 0.78 (0.14), 0.86 (0.08), 0.89 (0.08) and 0.71 (0.17), respectively. These results suggest that AI methods may assist less experienced residents or radiologists in liver MRI reporting of focal liver lesions.

Emergency Use and Efficacy of an Asynchronous Teledermatology System as a Novel Tool for Early Diagnosis of Skin Cancer during the First Wave of COVID-19 Pandemic.

Background: After the outbreak of the corona virus disease-19 (COVID-19) pandemic, teledermatology was implemented in the Hungarian public healthcare system for the first time. Our objective was to assess aggregated diagnostic agreements and to determine the effectiveness of an asynchronous teledermatology system for skin cancer screening. Methods: This retrospective single-center study included cases submitted for teledermatology consultation during the first wave of the COVID-19 pandemic. Follow-up of the patients was performed to collect the results of any subsequent personal examination. Results: 749 patients with 779 lesions were involved. 15 malignant melanomas (9.9%), 78 basal cell carcinomas (51.3%), 21 squamous cell carcinomas (13.8%), 7 other malignancies (4.6%) and 31 actinic keratoses (20.4%) were confirmed. 87 malignancies were diagnosed in the high-urgency group (42.2%), 49 malignancies in the moderate-urgency group (21.6%) and 16 malignancies in the low-urgency group (4.6%) (p < 0.0001). Agreement of malignancies was substantial for primary (86.3%; κ = 0.647) and aggregated diagnoses (85.3%; κ = 0.644). Agreement of total lesions was also substantial for primary (81.2%; κ = 0.769) and aggregated diagnoses (87.9%; κ = 0.754). Conclusions: Our findings showed that asynchronous teledermatology using a mobile phone application served as an accurate skin cancer screening system during the first wave of the COVID-19 pandemic.

Diagnosis of focal liver lesions with deep learning-based multi-channel analysis of hepatocyte-specific contrast-enhanced magnetic resonance imaging.

BACKGROUND: The nature of input data is an essential factor when training neural networks. Research concerning magnetic resonance imaging (MRI)-based diagnosis of liver tumors using deep learning has been rapidly advancing. Still, evidence to support the utilization of multi-dimensional and multi-parametric image data is lacking. Due to higher information content, three-dimensional input should presumably result in higher classification precision. Also, the differentiation between focal liver lesions (FLLs) can only be plausible with simultaneous analysis of multi-sequence MRI images. AIM: To compare diagnostic efficiency of two-dimensional (2D) and three-dimensional (3D)-densely connected convolutional neural networks (DenseNet) for FLLs on multi-sequence MRI. METHODS: We retrospectively collected T2-weighted, gadoxetate disodium-enhanced arterial phase, portal venous phase, and hepatobiliary phase MRI scans from patients with focal nodular hyperplasia (FNH), hepatocellular carcinomas (HCC) or liver metastases (MET). Our search identified 71 FNH, 69 HCC and 76 MET. After volume registration, the same three most representative axial slices from all sequences were combined into four-channel images to train the 2D-DenseNet264 network. Identical bounding boxes were selected on all scans and stacked into 4D volumes to train the 3D-DenseNet264 model. The test set consisted of 10-10-10 tumors. The performance of the models was compared using area under the receiver operating characteristic curve (AUROC), specificity, sensitivity, positive predictive values (PPV), negative predictive values (NPV), and f1 scores. RESULTS: The average AUC value of the 2D model (0.98) was slightly higher than that of the 3D model (0.94). Mean PPV, sensitivity, NPV, specificity and f1 scores (0.94, 0.93, 0.97, 0.97, and 0.93) of the 2D model were also superior to metrics of the 3D model (0.84, 0.83, 0.92, 0.92, and 0.83). The classification metrics of FNH were 0.91, 1.00, 1.00, 0.95, and 0.95 using the 2D and 0.90, 0.90, 0.95, 0.95, and 0.90 using the 3D models. The 2D and 3D networks' performance in the diagnosis of HCC were 1.00, 0.80, 0.91, 1.00, and 0.89 and 0.88, 0.70, 0.86, 0.95, and 0.78, respectively; while the evaluation of MET lesions resulted in 0.91, 1.00, 1.00, 0.95, and 0.95 and 0.75, 0.90, 0.94, 0.85, and 0.82 using the 2D and 3D networks, respectively. CONCLUSION: Both 2D and 3D-DenseNets can differentiate FNH, HCC and MET with good accuracy when trained on hepatocyte-specific contrast-enhanced multi-sequence MRI volumes.

Segmentation of knee cartilages in MR images with artificial intelligence / Térdporc szegmentálása MR-felvételekbol mesterséges intelligencia segítségével.

Összefoglaló. Bevezetés: A térdízületnek ultrafriss osteochondralis allograft segítségével történo részleges ortopédiai rekonstrukciója képalkotó vizsgálatokon alapuló pontos tervezést igényel, mely folyamatban a morfológia felismerésére képes mesterséges intelligencia nagy segítséget jelenthet. Célkituzés: Jelen kutatásunk célja a porc morfológiájának MR-felvételen történo felismerésére alkalmas mesterséges intelligencia kifejlesztése volt. Módszer: A feladatra legalkalmasabb MR-szekvencia meghatározása és 180 térd-MR-felvétel elkészítése után a mesterséges intelligencia tanításához manuálisan és félautomata szegmentálási módszerrel bejelölt porckontúrokkal tréninghalmazt hoztunk létre. A mély convolutiós neuralis hálózaton alapuló mesterséges intelligenciát ezekkel az adatokkal tanítottuk be. Eredmények: Munkánk eredménye, hogy a mesterséges intelligencia képes a meghatározott szekvenciájú MR-felvételen a porcnak a mutéti tervezéshez szükséges pontosságú bejelölésére, mely az elso lépés a gép által végzett mutéti tervezés felé. Következtetés: A választott technológia - a mesterséges intelligencia - alkalmasnak tunik a porc geometriájával kapcsolatos feladatok megoldására, ami széles köru alkalmazási lehetoséget teremt az ízületi terápiában. Orv Hetil. 2021; 162(9): 352-360. INTRODUCTION: The partial orthopedic reconstruction of the knee joint with an osteochondral allograft requires precise planning based on medical imaging reliant; an artificial intelligence capable of determining the morphology of the cartilage tissue can be of great help in such a planning. OBJECTIVE: We aimed to develop and train an artificial intelligence capable of determining the cartilage morphology in a knee joint based on an MR image. METHOD: After having determined the most appropriate MR sequence to use for this project and having acquired 180 knee MR images, we created the training set for the artificial intelligence by manually and semi-automatically segmenting the contours of the cartilage in the images. We then trained the neural network with this dataset. RESULTS: As a result of our work, the artificial intelligence is capable to determine the morphology of the cartilage tissue in the MR image to a level of accuracy that is sufficient for surgery planning, therefore we have made the first step towards machine-planned surgeries. CONCLUSION: The selected technology - artificial intelligence - seems capable of solving tasks related to cartilage geometry, creating a wide range of application opportunities in joint therapy. Orv Hetil. 2021; 162(9): 352-360.