Keywords: Pancreas, Image Reconstruction, AI, Super Resolution

Motivation: Abdominal diffusion-weighted imaging (DWI) plays a significant role in the detection and characterization of lesions. However, the spatial resolution of single-shot echo-planar imaging (ssh-EPI) readout is limited by the acquisition time.

Goal(s): To enhance the image quality and sharpness of abdominal ssh-EPI-DWI image using a prototype AI-based reconstruction technique (SuperRes).

Approach: We examined eight healthy volunteers using abdominal ssh-EPI-DWI, and the acquired data were reconstructed using both conventional Compressed SENSE and SuperRes. The image quality was assessed qualitatively and quantitatively.

Results: SuperRes demonstrated a significant improvement in the image quality and sharpness of both DWI and ADC map. 

Impact: The dedicated deep learning-based super-resolution technique enhanced the image quality and sharpness in abdominal ssh-EPI-DWI. Enhanced sharpness resulted in better delineation of structures, such as the pancreas. The improvement in image quality was demonstrated in both qualitative and quantitative assessments.

Keywords: Breast, Image Reconstruction

Motivation: DWI MRI is widely used in diagnosis and treatment evaluation of breast cancer but is prone to artifacts due to the breast's superficial location and large field of view (FOV).

Goal(s): To investigate the feasibility and performance of reduced-FOV FOCUS DWI and FOCUS DWI with deep learning-based reconstruction (DLR) for breast MRI in Asian patients with small breast volumes.

Approach: Both subjective and objective methods were used to compare the image quality of FOCUS DWI, FOCUS-DLR DWI and conventional DWI for breast cancer imaging.

Results: Our results demonstrated that FOCUS-DLR DWI showed improved image quality and higher image scores compared to conventional DWI.

Impact: FOCUS-DLR DWI enhances the visibility of lesion details, offering a novel approach to optimize breast MRI. This technique also holds promise for improving diffusion imaging in other regions of the human body, particularly small organs with surrounding tissue.

Keywords: Digestive, Diffusion/other diffusion imaging techniques

Motivation: Microsatellite instability (MSI) in esophagogastric junction adenocarcinoma (EGA) can serve as a predictor of sensitivity to immunotherapy and affect the prognosis. Predicting MSI preoperatively can enable personalized and precise treatment for EGA patients.

Goal(s): This study investigates the use of fast non-Gaussian diffusion-weighted imaging with deep learning-based reconstruction (DLRecon) to assess MSI in EGA.

Approach: We compared image quality between conventional scanning (CS) and DLRecon, calculated diffusion parameters, and assessed their ability to distinguish MSI status.

Results: DLRecon exhibited superior image quality and reduced scan time. Diffusion parameters effectively differentiated MSI status in EGA.

Impact: DLRecon non-Gaussian DWI significantly improved image quality and reduced acquisition time. Multiple diffusion parameters may serve as imaging markers, and their combination provides high diagnostic accuracy for discriminating MSI status in EGA.

Keywords: Prostate, Prostate

Motivation: Multiparametric prostate MRI is lengthy and costly, presenting a challenge for widespread implementation.

Goal(s): To develop an automated triage protocol using a deep learning classifier to discern, based on an abbreviated biparametric MR examination, between high-risk patients who would benefit from additional sequences and low-risk patients who would not.

Approach: A double-branched ResNet50 with 3D convolutions was trained on biparametric scans to predict the presence of clinically significant prostate cancer.

Results: The classifier achieved a sensitivity of 0.93 with 88% negative predictive value, indicating potential to reduce comprehensive MRI exams for those without clinically significant disease by 40%.

Impact: Our triage protocol has the potential to streamline prostate cancer screening by reducing the number of full mpMRI exams, thereby lowering healthcare costs. The classifier could pave the way for personalized, risk-adaptive screening protocols, allowing more precise and resource-efficient diagnostics.

Keywords: Cancer, Prostate, rostate cancer diagnosis, Gleason score prediction, Magnetic Resonance Imaging (MRI), Convolutional Neural Networks (CNN), Pathological upgrading detection, Lesion characterization, Clinical significance determination, Artificial Intelligence (AI) in medical imaging, Precision medicine, Patient management, Predictive modeling in oncology, Diagnostic precision, Risk assessment in prostate cancer, Biopsy accuracy, Medical imaging technologies, Machine learning in healthcare

Motivation: Patients with low risk (Gleason 6) MR visible prostate cancer on initial biopsy are frequently upgraded to aggressive higher risk (Gleason 7 or higher) cancer. Identifying this progression early is difficult.

Goal(s): To address this using a neural network trained with radiologist labels and whole mount histology of Gleason ≥7 cases to predict pathological upgrading in our cohort.

Approach: DecNet was applied to the Gleason 6 initial MRIs to assess if the model could retrospectively identify patients with higher grade disease.

Results: Our model had a sensitivity of 84.6% for lesions upgraded to Gleason 7, outperforming PSA density, lesion size and ADC values.

Impact: These results showcase the potential of our model in unveiling higher-grade prostate cancer within lesions initially diagnosed as lower grade on pathology.

Keywords: Prostate, Prostate

Motivation: The need to improve prostate cancer diagnosis through advanced understanding of lesion characteristics and reducing false positives led to this research.

Goal(s): To create a pioneering integrated system using deep learning, capable of accurately assessing the benignity or malignancy of prostate MRI images, whilst reducing labeling costs and enhancing the reliability of classifications.

Approach: The approach involves training a convolutional network with multi-parametric MRI images, incorporating credibility analysis to provide visually interpretable prostate cancer prediction results and reject low-credibility predictions. 

Results: The results showed improved reliability and efficacy, with the model discarding low-credibility predictions, thus mitigating potential risks associated with prediction failures.

Impact: This study equips clinical practitioners with the ability to comprehend the decision-making process of the CAD system and manage the output results through an intuitive display. This results enhance diagnostic accuracy, potentially impacting clinicians' decision-making and patient outcomes.

Keywords: Pancreas, Pancreas

Motivation: Cystic fluid appears hyperintense via T2WI, the most sensitive detection method and T2WI is a conventional sequence. However, distinguishing pancreatic MCNs from SCNs using T2WI is difficult because both neoplasms appear as hyperintense lesions, especially when both are unilocular.

Goal(s): We aimed to develop and validate a T2WI radiomics nomogram for the differentiation of SCNs from MCNs.

Approach: A radiomics model that was included clinical characteristics, MRI characteristics, and T2WI rad-scores for differentiating MCNs from SCN.

Results: We developed and validated a T2WI radiomics nomogram that functions as a non-invasive and convenient tool for preoperatively predicting the presence of SCNs and MCNs.

Impact: The tool has the potential to help clinicians identify patients requiring surveillance or surgery.

Keywords: Kidney, Radiomics

Motivation: Kidney cancer is often diagnosed as either clear cell renal carcinoma (ccRCC) or non-clear cell renal carcinoma (non-ccRCC) to determine treatment recommendations. Additionally, many patients with kidney cancer cannot receive contrast medium due to renal function disorders.

Goal(s): for the distinction of ccRCC from other types of RCC without contrast medium administration

Approach: A model using automated machine learning (AutoML) based on radiomics features 

Results: Our results indicate that the best model from the AutoML process demonstrated a mean sensitivity of 0.819 and a mean specificity of 0.729 in distinguishing between ccRCC and non-ccRCC.

Impact: To demonstrated that the TPOP-radiomics-based classification model can effectively discriminate between ccRCC and non-ccRCC using MRI without the need for contrast medium.

Keywords: Digestive, Radiomics, Crohn’s disease; Fibrosis; MR Enterography

Motivation: Fibrostenosis is a severe complication of Crohn's disease that significantly impacts patients' quality of life. Currently, there are no effective medical interventions for severe intestinal fibrosis

Goal(s): We aimed to develop MRI-based radiomic models to improve the diagnostic accuracy of radiologists in characterizing intestinal fibrosis in patients with Crohn’s disease.

Approach: Radiomic models based on different MR sequence combinations were developed and validated in an independent test cohort. 

Results: The radiologists’ interpretation aided by MRI-radiomics outperformed visual interpretation in characterizing intestinal fibrosis in Crohn's disease (AUC=0.86-0.93 vs. AUC=0.63-0.77).

Impact: The utilization of MRI-based radiomic models significantly enhances the diagnostic accuracy of radiologists in characterizing intestinal fibrosis.

Keywords: Fetal, Data Processing

Motivation: The cerebral cortex of the fetus is undergoing intricate development. Abnormal cortical development may potentially alter brain function. However the methods for processing fetal MRI images, especially cortical reconstruction, are still far behind those used for adults.

Goal(s): To develop an accurate cortical surface reconstruction method and perform morphological calculations for fetal MRI.

Approach: Trustworthy AI segmentation and refined Freesurfer were used to process T₂-weighted fetal brain image. 

Results: The proposed mothod (FetalSurfer) allows for trustworthy reconstruction of the cortical surfaces of the fetal brain and calculation of indicators to measure the morphology of fetal development.

Impact: FetalSurfer implements a novel fetal cerebral cortex reconstruction method without manual refinement by professional doctors, filling the gap of fetal image processing methods. Calculated indicators as curvature, thickness and sulcal depth can be used to perform morphological analysis.

Keywords: Hepatobiliary, Radiomics

Motivation: To the best of our knowledge, our study is the first to non-invasively assess methyltransferase-like 3 (METTL3) expression in hepatocellular carcinoma (HCC) using multi-sequence MRI-based radiomics.

Goal(s): To develop a multi-region radiomics-based predictive model using multi-sequence MRI to assess METTL3 expression in HCC.

Approach: Three Models (Tumor, Tumor-Expand5, Tumor-Expand10) were constructed and evaluated.

Results: The Tumor-Expand5 model showed the highest efficacy with an AUC of 0.71 in the test set, outperforming both the Tumor and Tumor-Expand10 models. Multi-sequence MRI-based radiomics models hold the potential for preoperatively assessing METTL3 expression in HCC, aiding clinical decision-making.

Impact: This study introduces a novel multi-region radiomics-based model for predicting METTL3 expression in HCC using multi-sequence MRI. The results demonstrate the potential of radiomics, with an emphasis on the Tumor-Expand5 model, highlighting its promise for enhancing clinical decision-making in HCC.

Keywords: Digestive, Cancer, Rctal; Prognosis

Motivation: The prognostic value and importance of peritumoral radiomics remain underexplored in locally advanced rectal cancer (LARC).

Goal(s): To investigate the prognostic significance of peritumoral versus intratumoral radiomic features in LARC.

Approach: In a retrospective cohort of 409 patients with LARC, we extracted intratumoral and peritumoral radiomic features from pretreatment high-resolution small-field-of-view T2-weighted images. Various prognostic models incorporating clinicopathological and radiomic data were constructed and compared. Variable importance was analyzed.

Results: Peritumoral features demonstrated equivalent or superior prognostic value than intratumoral features, significantly enhanced accuracy of models relying on intratumoral or intratumoral-clinicpathological features. One peritumoral feature emerged as the leading predictor.

Impact: Peritumoral radiomics provides equal or even greater prognostic value compared to intratumoral radiomics, promising to enhance accuracy in prognostic estimation for locally advanced rectal cancer and thus facilitating personalized treatment strategies.

Keywords: Hepatobiliary, Low-Field MRI, Abdomen, deep-learning reconstruction

Motivation: Deep-learning reconstruction may overcome two shortcomings of 0.55T, low SNR and extended scan time, without compromising lesion conspicuity.

Goal(s): To demonstrate that image quality and SNR of deep-learning reconstructed 0.55T images are at least similar to 1.5T/3T images, while maintaining visibility of pathologies.

Approach: 23 patients imaged at 0.55T using standard and deep-learning HASTE and DWI. Three radiologists rated IQ and SNR at 0.55T and HF. Pathologies were evaluated in deep-learning images.

Results: Deep-learning reconstructed HASTE and DWI 0.55T images were of same or better quality and SNR than 1.5T/3T images. All pathologies were visible on deep-learning 0.55T images. DL reduced HASTE scan-time.

Impact: Deep-learning reconstruction algorithms of select sequences at 0.55T can help overcome low SNR and extended scan times of current 0.55T abdominal imaging, making it comparable or superior to standard-of-care 1.5/3T, thereby expanding global use of a more accessible MRI system.

Keywords: Kidney, Elastography, Magnetic resonance elastography, chronic kidney disease, renal stiffness

Motivation: The incorporation of automated kidney segmentation technology with multi-frequency MRE represents a novel approach in evaluating kidney diseases.

Goal(s): To explore the reliability of multi-frequency MRE combined with an automatic segmentation method and its diagnostic potential for CKD patients.

Approach: Constructed an automatic kidney segmentation model based on the nnU-Net network and measured the renal stiffness of MRE, employing T-tests, ROC curves, and Spearman correlation for data analysis.

Results: The incorporation of an automatic kidney segmentation model and multi-frequency MRE shows promise in effectively evaluating and monitoring kidney fibrosis.

Impact: The incorporation of automated kidney segmentation and MRE presents a new tool for reliably evaluating and monitoring kidney diseases, providing potential advancements in non-invasive diagnoses.

Keywords: Digestive, Biliary

Motivation: Differentiating CBD dilatation with CT alone is challenging, often necessitating MRCP(Magnetic Resonance Cholangiopancreatography). Yet, patients forego MRCP due to cost and time constraints. Hence, predicting CBD dilatation using CT is vital for diagnosis.

Goal(s): Developing deep neural networks to assess CBD dilatation only with CT data.

Approach: Cycle-GAN and 3D VGG Networks predicted CBD dilatation, where Cycle-GAN generated synthetic MRCP from CT and 3D VGG Network predicted dilatation using this synthetic data.

Results: The network trained with synthetic MRCP data predicted CBD dilatation with an AUROC of 0.7231, 30% improvement over using CT data alone, enabling CT-only diagnosis.

Impact: This study introduces a transformative solution for CBD dilatation diagnosis, enabling assessments using Only-CT data from Cycle-GAN and 3D VGG Network. Achieving a 30% improvement in AUROC, it enables reliable CT-only diagnoses, overcoming scarce MRCP data and improve patient care.

Keywords: Liver, Liver, Foundation model, Segmentation

Motivation: Accurate hepatic vessel segmentation can help to identify and avoid critical blood vessels during liver tumor ablation or resection. Existing methods are not accessible to most medical institutes, leading to questionable clinical relevance.

Goal(s): we present a handy foundation model-based hepatic vessel segmentation approach crafted for straightforward integration into clinical applications.

Approach: We employ a parameter-efficient few-shot learning strategy to fine-tune the foundation model, thereby enabling it to achieve competitive hepatic vessel segmentation performance with training on only five cases.

Results: The proposed method is effective and easy-to-access, and it has the potential for a substantial impact on clinical practice.

Impact: Existing hepatic vessel segmentation methods are not accessible to most medical institutes, leading to questionable clinical relevance. We present a clinically practical foundation model-based approach that achieves competitive performance with training on only five cases.