Deep Learning: A Primer for Radiologists

Prompt Engineering for Large Language Models in Interventional Radiology

Prompt engineering plays a crucial role in optimizing artificial intelligence (AI) and large language model (LLM) outputs by refining input structure, a key factor in medical applications where precision and reliability are paramount. This Clinical Perspective provides an overview of prompt engineering techniques and their relevance to interventional radiology (IR). It explores key strategies, including zero-shot, one- or few-shot, chain-of-thought, tree-of-thought, self-consistency, and directional stimulus prompting, demonstrating their application in IR-specific contexts. Practical examples illustrate how these techniques can be effectively structured for workplace and clinical use. Additionally, the article discusses best practices for designing effective prompts and addresses challenges in the clinical use of generative AI, including data privacy and regulatory concerns. It concludes with an outlook on the future of generative AI in IR, highlighting advances including retrieval-augmented generation, domain-specific LLMs, and multimodal models.

Advancing periodontal diagnosis: harnessing advanced artificial intelligence for patterns of periodontal bone loss in cone-beam computed tomography

OBJECTIVES

The current study aimed to automatically detect tooth presence, tooth numbering, and types of periodontal bone defects from cone-beam CT (CBCT) images using a segmentation method with an advanced artificial intelligence (AI) algorithm.

METHODS

This study utilized a dataset of CBCT volumes collected from 502 individual subjects. Initially, 250 CBCT volumes were used for automatic tooth segmentation and numbering. Subsequently, CBCT volumes from 251 patients diagnosed with periodontal disease were employed to train an AI system to identify various periodontal bone defects using a segmentation method in web-based labelling software. In the third stage, CBCT images from 251 periodontally healthy subjects were combined with images from 251 periodontally diseased subjects to develop an AI model capable of automatically classifying patients as either periodontally healthy or periodontally diseased. Statistical evaluation included receiver operating characteristic curve analysis and confusion matrix model.

RESULTS

The area under the receiver operating characteristic curve (AUC) values for the models developed to segment teeth, total alveolar bone loss, supra-bony defects, infra-bony defects, perio-endo lesions, buccal defects, and furcation defects were 0.9594, 0.8499, 0.5052, 0.5613 (with cropping, AUC: 0.7488), 0.8893, 0.6780 (with cropping, AUC: 0.7592), and 0.6332 (with cropping, AUC: 0.8087), respectively. Additionally, the classification CNN model achieved an accuracy of 80% for healthy individuals and 76% for unhealthy individuals.

CONCLUSIONS

This study employed AI models on CBCT images to automatically detect tooth presence, numbering, and various periodontal bone defects, achieving high accuracy and demonstrating potential for enhancing dental diagnostics and patient care.

Perfusion estimation from dynamic non-contrast computed tomography using self-supervised learning and a physics-inspired U-net transformer architecture

PURPOSE

Pulmonary perfusion imaging is a key lung health indicator with clinical utility as a diagnostic and treatment planning tool. However, current nuclear medicine modalities face challenges like low spatial resolution and long acquisition times which limit clinical utility to non-emergency settings and often placing extra financial burden on the patient. This study introduces a novel deep learning approach to predict perfusion imaging from non-contrast inhale and exhale computed tomography scans (IE-CT).

METHODS

We developed a U-Net Transformer architecture modified for Siamese IE-CT inputs, integrating insights from physical models and utilizing a self-supervised learning strategy tailored for lung function prediction. We aggregated 523 IE-CT images from nine different 4DCT imaging datasets for self-supervised training, aiming to learn a low-dimensional IE-CT feature space by reconstructing image volumes from random data augmentations. Supervised training for perfusion prediction used this feature space and transfer learning on a cohort of 44 patients who had both IE-CT and single-photon emission CT (SPECT/CT) perfusion scans.

RESULTS

Testing with random bootstrapping, we estimated the mean and standard deviation of the spatial Spearman correlation between our predictions and the ground truth (SPECT perfusion) to be 0.742 ± 0.037, with a mean median correlation of 0.792 ± 0.036. These results represent a new state-of-the-art accuracy for predicting perfusion imaging from non-contrast CT.

CONCLUSION

Our approach combines low-dimensional feature representations of both inhale and exhale images into a deep learning model, aligning with previous physical modeling methods for characterizing perfusion from IE-CT. This likely contributes to the high spatial correlation with ground truth. With further development, our method could provide faster and more accurate lung function imaging, potentially expanding its clinical applications beyond what is currently possible with nuclear medicine.

Large multimodality model fine-tuned for detecting breast and esophageal carcinomas on CT: a preliminary study

PURPOSE

This study aimed to develop a large multimodality model (LMM) that can detect breast and esophageal carcinomas on chest contrast-enhanced CT.

MATERIALS AND METHODS

In this retrospective study, CT images of 401 (age, 62.9 ± 12.9 years; 169 males), 51 (age, 65.5 ± 11.6 years; 23 males), and 120 (age, 64.6 ± 14.2 years; 60 males) patients were used in the training, validation, and test phases. The numbers of CT images with breast carcinoma, esophageal carcinoma, and no lesion were 927, 2180, and 2087; 80, 233, and 270; and 184, 246, and 6919 for the training, validation, and test datasets, respectively. The LMM was fine-tuned using CT images as input and text data ("suspicious of breast carcinoma"/ "suspicious of esophageal carcinoma"/ "no lesion") as reference data on a desktop computer equipped with a single graphic processing unit. Because of the random nature of the training process, supervised learning was performed 10 times. The performance of the best performing model on the validation dataset was further tested using the time-independent test dataset. The detection performance was evaluated by calculating the area under the receiver operating characteristic curve (AUC).

RESULTS

The sensitivities of the fine-tuned LMM for detecting breast and esophageal carcinomas in the test dataset were 0.929 and 0.951, respectively. The diagnostic performance of the fine-tuned LMM for detecting breast and esophageal carcinomas was high, with AUCs of 0.890 (95%CI 0.871-0.909) and 0.880 (95%CI 0.865-0.894), respectively.

CONCLUSIONS

The fine-tuned LMM could detect both breast and esophageal carcinomas on chest contrast-enhanced CT with high diagnostic performance. Usefulness of large multimodality models in chest cancer imaging has not been assessed so far. The fine-tuned large multimodality model could detect breast and esophageal carcinomas with high diagnostic performance (area under the receiver operating characteristic curve of 0.890 and 0.880, respectively).

Artificial Intelligence in Gastrointestinal Imaging: Advances and Applications

While artificial intelligence (AI) has shown considerable progress in many areas of medical imaging, applications in abdominal imaging, particularly for the gastrointestinal (GI) system, have notably lagged behind advancements in other body regions. This article reviews foundational concepts in AI and highlights examples of AI applications in GI tract imaging. The discussion on AI applications includes acute & emergent GI imaging, inflammatory bowel disease, oncology, and other miscellaneous applications. It concludes with a discussion of important considerations for implementing AI tools in clinical practice, and steps we can take to accelerate future developments in the field.

Association of long-term treatment outcomes with changes in PET/MRI characteristics and the type of early treatment response during concurrent radiochemotherapy in patients with locally advanced cervical cancer

PURPOSE

We aimed to find predictive tumour characteristics as detected by interim positron-emission tomography/magnetic resonance imaging (PET/MRI) in cervical cancer patients. We also investigated the type of interim response. Furthermore, we compared the investigated parameters with disease-free (DFS) and overall survival (OS) outcomes.

METHODS

We evaluated 108 patients treated between August 2015 and January 2023 with external-beam radiotherapy (EBRT) and image-guided adaptive brachytherapy (IGABT) who had undergone pretreatment staging, subsequent mid-treatment evaluation after completed EBRT and definitive restaging 3 months after completing the whole treatment using PET/MRI. Patients were then divided into two groups based on the RECIST and PERCIST criteria: responders (achieving complete metabolic response, CMR) and non-responders (non-CMR). These two groups were compared using selected parameters obtained at pre-PET/MRI and mid-PET/MRI. The early response to treatment as evaluated by mid-PET/MRI was categorized into three types: interim complete metabolic response, interim nodal response and interim nodal persistence.

RESULTS

Mid-TLG‑S (the sum of total lesion glycolysis for the primary tumour plus pelvic and para-aortic lymph nodes) parameter showed the best discriminatory ability for predicting non-CMR. The second factor with significant discriminatory ability was mid-MTV‑S (the sum of the metabolic tumour volume of the primary tumour plus pelvic and para-aortic lymph nodes). The strongest factor, mid-TLG‑S, showed a sensitivity of 40% and a specificity of 90% at a threshold value of 70. We found a statistically significant association of DFS and OS with the following parameters: number of chemotherapy cycles, early response type and CMR vs. non-CMR.

CONCLUSION

We were able to identify thresholds for selected parameters that can be used to identify patients who are more likely to have worse DFS and OS. The type of early response during concurrent chemoradiotherapy (CCRT) was also significantly associated with DFS and OS. These aspects represent an important contribution to the possible stratification of patients for subsequent individualised adjuvant treatment.

Multivariable prognostic models for post-hepatectomy liver failure: An updated systematic review

BACKGROUND

Partial hepatectomy continues to be the primary treatment approach for liver tumors, and post-hepatectomy liver failure (PHLF) remains the most critical life-threatening complication following surgery.

AIM

To comprehensively review the PHLF prognostic models developed in recent years and objectively assess the risk of bias in these models.

METHODS

This review followed the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline. Three databases were searched from November 2019 to December 2022, and references as well as cited literature in all included studies were manually screened in March 2023. Based on the defined inclusion criteria, articles on PHLF prognostic models were selected, and data from all included articles were extracted by two independent reviewers. The PROBAST was used to evaluate the quality of each included article.

RESULTS

A total of thirty-four studies met the eligibility criteria and were included in the analysis. Nearly all of the models (32/34, 94.1%) were developed and validated exclusively using private data sources. Predictive variables were categorized into five distinct types, with the majority of studies (32/34, 94.1%) utilizing multiple types of data. The area under the curve for the training models included ranged from 0.697 to 0.956. Analytical issues resulted in a high risk of bias across all studies included.

CONCLUSION

The validation performance of the existing models was substantially lower compared to the development models. All included studies were evaluated as having a high risk of bias, primarily due to issues within the analytical domain. The progression of modeling technology, particularly in artificial intelligence modeling, necessitates the use of suitable quality assessment tools.

Super-resolution deep learning reconstruction to evaluate lumbar spinal stenosis status on magnetic resonance myelography

PURPOSE

To investigate whether super-resolution deep learning reconstruction (SR-DLR) of MR myelography-aided evaluations of lumbar spinal stenosis.

MATERIAL AND METHODS

In this retrospective study, lumbar MR myelography of 40 patients (16 males and 24 females; mean age, 59.4 ± 31.8 years) were analyzed. Using the MR imaging data, MR myelography was separately reconstructed via SR-DLR, deep learning reconstruction (DLR), and conventional zero-filling interpolation (ZIP). Three radiologists, blinded to patient background data and MR reconstruction information, independently evaluated the image sets in terms of the following items: the numbers of levels affected by lumbar spinal stenosis; and cauda equina depiction, sharpness, noise, artifacts, and overall image quality.

RESULTS

The median interobserver agreement in terms of the numbers of lumbar spinal stenosis levels were 0.819, 0.735, and 0.729 for SR-DLR, DLR, and ZIP images, respectively. The imaging quality of the cauda equina, and image sharpness, noise, and overall quality on SR-DLR images were significantly better than those on DLR and ZIP images, as rated by all readers (p < 0.001, Wilcoxon signed-rank test). No significant differences were observed for artifacts on SR-DLR against DLR and ZIP.

CONCLUSIONS

SR-DLR improved the image quality of lumbar MR myelographs compared to DLR and ZIP, and was associated with better interobserver agreement during assessment of lumbar spinal stenosis status.

A deep learning-based calculation system for plaque stenosis severity on common carotid artery of ultrasound images

ObjectivesAssessment of plaque stenosis severity allows better management of carotid source of stroke. Our objective is to create a deep learning (DL) model to segment carotid intima-media thickness and plaque and further automatically calculate plaque stenosis severity on common carotid artery (CCA) transverse section ultrasound images.MethodsThree hundred and ninety images from 376 individuals were used to train (235/390, 60%), validate (39/390, 10%), and test (116/390, 30%) on a newly proposed CANet model. We also evaluated the model on an external test set of 115 individuals with 122 images acquired from another hospital. Comparative studies were conducted between our CANet model with four state-of-the-art DL models and two experienced sonographers to re-evaluate the present model's performance.ResultsOn the internal test set, our CANet model outperformed the four comparative models with Dice values of 95.22% versus 90.15%, 87.48%, 90.22%, and 91.56% on lumen-intima (LI) borders and 96.27% versus 91.40%, 88.94%, 91.19%, and 92.88% on media-adventitia (MA) borders. On the external test set, our model still produced excellent results with a Dice value of 92.41%. Good consistency of stenosis severity calculation was observed between CANet model and experienced sonographers, with Intraclass Correlation Coefficient (ICC) of 0.927 and 0.702, Pearson's Correlation Coefficient of 0.928 and 0.704 on internal and external test set, respectively.ConclusionsOur CANet model achieved excellent performance in the segmentation of carotid IMT and plaques as well as automated calculation of stenosis severity.

Role of artificial intelligence in magnetic resonance imaging-based detection of temporomandibular joint disorder: a systematic review

This systematic review aimed to evaluate the application of artificial intelligence (AI) in the identification of temporomandibular joint (TMJ) disc position in normal or temporomandibular joint disorder (TMD) individuals using magnetic resonance imaging (MRI). Database search was done in Pub med, Google scholar, Semantic scholar and Cochrane for studies on AI application to detect TMJ disc position in MRI till September 2023 adhering PRISMA guidelines. Data extraction included number of patients, number of TMJ/MRI, AI algorithm and performance metrics. Risk of bias was done with modified PROBAST tool. Seven studies were included (deep learning = 6, machine learning = 1). Sensitivity values (n = 7) ranged from 0.735 to 1, while specificity values (n = 4) ranged from 0.68 to 0.961. AI achieves accuracy levels exceeding 83%. MobileNetV2 and ResNet have revealed better performance metrics. Machine learning demonstrated the lowest accuracy 74.2%. Risk of bias was low (n = 6) and high (n = 1). Deep learning models showed reliable performance metrics for AI based detection of temporomandibular joint disc position in MRI. Future research is warranted with better standardisation of design and consistent reporting.

Comparison of conventional diffusion-weighted imaging and multiplexed sensitivity-encoding combined with deep learning-based reconstruction in breast magnetic resonance imaging

PURPOSE

To evaluate the feasibility of multiplexed sensitivity-encoding (MUSE) with deep learning-based reconstruction (DLR) for breast imaging in comparison with conventional diffusion-weighted imaging (DWI) and MUSE alone.

METHODS

This study was conducted using conventional single-shot DWI and MUSE data of female participants who underwent breast magnetic resonance imaging (MRI) from June to December 2023. The k-space data in MUSE were reconstructed using both conventional reconstruction and DLR. Two experienced radiologists conducted quantitative analyses of DWI, MUSE, and MUSE-DLR images by obtaining the signal-to-noise ratio (SNR) and the contrast-to-noise ratio (CNR) of lesions and normal tissue and qualitative analyses by using a 5-point Likert scale to assess the image quality. Inter-reader agreement was assessed using the intraclass correlation coefficient (ICC). Image scores, SNR, CNR, and apparent diffusion coefficient (ADC) measurements among the three sequences were compared using the Friedman test, with significance defined at P < 0.05.

RESULTS

In evaluations of the images of 51 female participants using the three sequences, the two radiologists exhibited good agreement (ICC = 0.540-1.000, P < 0.05). MUSE-DLR showed significantly better SNR than MUSE (P < 0.001), while the ADC values within lesions and tissues did not differ significantly among the three sequences (P = 0.924, P = 0.636, respectively). In the subjective assessments, MUSE and MUSE-DLR scored significantly higher than conventional DWI in overall image quality, geometric distortion and axillary lymph node (P < 0.001).

CONCLUSION

In comparison with conventional DWI, MUSE-DLR yielded improved image quality with only a slightly longer acquisition time.

Multimodal deep learning fusion of ultrafast-DCE MRI and clinical information for breast lesion classification

BACKGROUND

Breast cancer is the most common cancer worldwide, and magnetic resonance imaging (MRI) constitutes a very sensitive technique for invasive cancer detection. When reviewing breast MRI examination, clinical radiologists rely on multimodal information, composed of imaging data but also information not present in the images such as clinical information. Most machine learning (ML) approaches are not well suited for multimodal data. However, attention-based architectures, such as Transformers, are flexible and therefore good candidates for integrating multimodal data.

PURPOSE

The aim of this study was to develop and evaluate a novel multimodal deep learning (DL) model combining ultrafast dynamic contrast-enhanced (UF-DCE) MRI images, lesion characteristics and clinical information for breast lesion classification.

MATERIALS AND METHODS

From 2019 to 2023, UF-DCE breast images and radiology reports of 240 patients were retrospectively collected from a single clinical center and annotated. Imaging data were constituted of volumes of interest (VOI) extracted around segmented lesions. Non-imaging data were constituted of both clinical (categorical) and geometrical (scalar) data. Clinical data were extracted from annotated reports and were associated to their corresponding lesions. We compared the diagnostic performances of traditional ML methods for non-imaging data, an image model based on the DL architecture, and a novel Transformer-based architecture, the Multimodal Sieve Transformer with Vision Transformer encoder (MMST-V).

RESULTS

The final dataset included 987 lesions (280 benign, 121 malignant lesions, and 586 benign lymph nodes) and 1081 reports. For classification with non-imaging data, scalar data had a greater influence on performances of lesion classification (Area under the receiver operating characteristic curve (AUROC) = 0.875 ± 0.042) than categorical data (AUROC = 0.680 ± 0.060). MMST-V achieved better performances (AUROC = 0.928 ± 0.027) than classification based on non-imaging data (AUROC = 0.900 ± 0.045), and imaging data only (AUROC = 0.863 ± 0.025).

CONCLUSION

The proposed MMST-V is an adaptative approach that can consider redundant information provided by multimodal information. It demonstrated better performances than unimodal methods. Results highlight that the combination of clinical patient data and detailed lesion information as additional clinical knowledge enhances the diagnostic performances of UF-DCE breast MRI.

Artificial intelligence in emergency neuroradiology: Current applications and perspectives

Emergency neuroradiology provides rapid diagnostic decision-making and guidance for management for a wide range of acute conditions involving the brain, head and neck, and spine. This narrative review aims at providing an up-to-date discussion about the state of the art of applications of artificial intelligence in emergency neuroradiology, which have substantially expanded in depth and scope in the past few years. A detailed analysis of machine learning and deep learning algorithms in several tasks related to acute ischemic stroke involving various imaging modalities, including a description of existing commercial products, is provided. The applications of artificial intelligence in acute intracranial hemorrhage and other vascular pathologies such as intracranial aneurysm and arteriovenous malformation are discussed. Other areas of emergency neuroradiology including infection, fracture, cord compression, and pediatric imaging are further discussed in turn. Based on these discussions, this article offers insight into practical considerations regarding the applications of artificial intelligence in emergency neuroradiology, calling for more development driven by clinical needs, attention to pediatric neuroimaging, and analysis of real-world performance.

Artificial intelligence for abdominopelvic trauma imaging: trends, gaps, and future directions

Abdominopelvic trauma is a major cause of morbidity and mortality, typically resulting from high-energy mechanisms such as motor vehicle collisions and penetrating injuries. Admission abdominopelvic trauma CT, performed either selectively or as part of a whole-body CT protocol, has become the workhorse screening and surgical planning modality due to improvements in speed and image quality. Radiography remains an essential element of the secondary trauma survey, and Focused Assessment with Sonography for Trauma (FAST) scanning has added value for quick assessment of non-compressible hemorrhage in hemodynamically unstable patients. Complex and severe polytrauma cases often delay radiology report turnaround times, which can potentially impede urgent clinical decision-making. Artificial intelligence (AI) computer-aided detection and diagnosis (CAD) offers promising solutions for enhanced diagnostic efficiency and accuracy in abdominopelvic trauma imaging. Although commercial AI tools for abdominopelvic trauma are currently available for only a few use cases, the literature reveals robust research and development (R&D) of prototype tools. Multiscale convolutional neural networks (CNNs) and transformer-based models are capable of detecting and quantifying solid organ injuries, fractures, and hemorrhage with a high degree of precision. Further, generalist foundation models such as multimodal vision-language models (VLMs) can be adapted and fine-tuned using imaging, clinical, and text data for a range of tasks, including detection, quantitative visualization, prognostication, and report auto-generation. Despite their promise, for most use cases in abdominopelvic trauma, AI CAD tools remain in the pilot stages of technology readiness, with persistent challenges related to data availability; the need for open-access PACS compatible software pipelines for pre-clinical shadow-testing; lack of well-designed multi-institutional validation studies; and regulatory hurdles. This narrative review provides a snapshot of the current state of AI in abdominopelvic trauma, examining existing commercial tools; research and development throughout the technology readiness pipeline; and future directions in this domain.

Imaging genomics of cancer: a bibliometric analysis and review

BACKGROUND

Imaging genomics is a burgeoning field that seeks to connections between medical imaging and genomic features. It has been widely applied to explore heterogeneity and predict responsiveness and disease progression in cancer. This review aims to assess current applications and advancements of imaging genomics in cancer.

METHODS

Literature on imaging genomics in cancer was retrieved and selected from PubMed, Web of Science, and Embase before July 2024. Detail information of articles, such as systems and imaging features, were extracted and analyzed. Citation information was extracted from Web of Science and Scopus. Additionally, a bibliometric analysis of the included studies was conducted using the Bibliometrix R package and VOSviewer.

RESULTS

A total of 370 articles were included in the study. The annual growth rate of articles on imaging genomics in cancer is 24.88%. China (133) and the USA (107) were the most productive countries. The top 2 keywords plus were "survival" and "classification". The current research mainly focuses on the central nervous system (121) and the genitourinary system (110, including 44 breast cancer articles). Despite different systems utilizing different imaging modalities, more than half of the studies in each system employed radiomics features.

CONCLUSIONS

Publication databases provide data support for imaging genomics research. The development of artificial intelligence algorithms, especially in feature extraction and model construction, has significantly advanced this field. It is conducive to enhancing the related-models' interpretability. Nonetheless, challenges such as the sample size and the standardization of feature extraction and model construction must overcome. And the research trends revealed in this study will guide the development of imaging genomics in the future and contribute to more accurate cancer diagnosis and treatment in the clinic.

Artificial intelligence in cardiovascular magnetic resonance imaging

Artificial intelligence is rapidly evolving and its possibilities are endless. Its primary applications in cardiac magnetic resonance imaging have focused on: image acquisition (in terms of acceleration and quality improvement); segmentation (in terms of saving time and reproducibility); tissue characterisation (including radiomic techniques and the non-contrast assessment of myocardial fibrosis); automatic diagnosis; and prognostic stratification. The aim of this article is to attempt to provide an overview of the current situation as preparation for the significant changes currently underway or imminent in the very near future.

Assessment of perceived realism in AI-generated synthetic spine fracture CT images

BackgroundDeep learning-based decision support systems require synthetic images generated by adversarial networks, which require clinical evaluation to ensure their quality.ObjectiveThe study evaluates perceived realism of high-dimension synthetic spine fracture CT images generated Progressive Growing Generative Adversarial Networks (PGGANs).Method: The study used 2820 spine fracture CT images from 456 patients to train an PGGAN model. The model synthesized images up to 512 × 512 pixels, and the realism of the generated images was assessed using Visual Turing Tests and Fracture Identification Test. Three spine surgeons evaluated the images, and clinical evaluation results were statistically analysed.Result: Spine surgeons have an average prediction accuracy of nearly 50% during clinical evaluations, indicating difficulty in distinguishing between real and generated images. The accuracy varies for different dimensions, with synthetic images being more realistic, especially in 512 × 512-dimension images. During FIT, among 16 generated images of each fracture type, 13-15 images were correctly identified, indicating images are more realistic and clearly depict fracture lines in 512 × 512 dimensions.ConclusionThe study reveals that AI-based PGGAN can generate realistic synthetic spine fracture CT images up to 512 × 512 pixels, making them difficult to distinguish from real images, and improving the automatic spine fracture type detection system.

Automatic Detection of Radiographic Alveolar Bone Loss in Bitewing and Periapical Intraoral Radiographs Using Deep Learning Technology: A Preliminary Evaluation

Background/Objective: Periodontal disease is a prevalent inflammatory condition affecting the supporting structures of teeth, with radiographic bone loss (RBL) being a critical diagnostic marker. The accurate and consistent evaluation of RBL is essential for the staging and grading of periodontitis, as outlined by the 2017 AAP/EFP Classification. Advanced tools such as deep learning (DL) technology, including Denti.AI, an FDA-cleared software utilizing convolutional neural networks (CNNs), offer the potential for enhancing diagnostic accuracy. This study evaluated the diagnostic accuracy of Denti.AI for detecting RBL in intraoral radiographs. Methods: A dataset of 39 intraoral radiographs (22 periapical and 17 bitewing), covering 316 tooth surfaces (123 periapical and 193 bitewing), was selected from a de-identified pool of 500 radiographs provided by Denti.AI. RBL was assessed using the 2017 AAP/EFP Classification. A consensus panel of three board-certified dental specialists served as the reference standard. Performance metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and mean absolute error (MAE), were calculated. Results: For periapical radiographs, Denti.AI achieved a sensitivity of 76%, specificity of 86%, PPV of 83%, NPV of 80%, and accuracy of 81%, with an MAE of 0.046%. For bitewing radiographs, sensitivity was 65%, specificity was 90%, PPV was 88%, NPV was 70%, and accuracy was 76%, with an MAE of 0.499 mm. Conclusions: Denti.AI demonstrated clinically acceptable performance in detecting RBL and shows potential as an adjunctive diagnostic tool, supporting clinical decision-making. While performance was robust for periapical radiographs, further optimization may enhance its accuracy for bitewing radiographs.

An Assessment of Deep Learning's Impact on General Dentists' Ability to Detect Alveolar Bone Loss in 2D Intraoral Radiographs

Background/Objective: Deep learning (DL) technology has shown potential in enhancing diagnostic accuracy in dentomaxillofacial radiology, particularly for detecting carious lesions, apical lesions, and periodontal bone loss. However, its effect on general dentists' ability to detect radiographic bone loss (RBL) in clinical practice remains unclear. This study investigates the impact of the Denti.AI DL technology on general dentists' ability to identify bone loss in intraoral radiographs, addressing this gap in the literature. Methods: Ten dentists from the university's dental clinics independently assessed 26 intraoral radiographs (periapical and bitewing) for bone loss using a Likert scale probability index with and without DL assistance. The participants viewed images on identical monitors with controlled lighting. This study generated 3940 data points for analysis. The statistical analyses included receiver operating characteristic (ROC) curves, area under the curve (AUC), and ANOVA tests. Results: Most dentists showed minor improvement in detecting bone loss on periapical radiographs when using DL. For bitewing radiographs, only a few dentists showed minor improvement. Overall, the difference in diagnostic accuracy between evaluations with and without DL was minimal (0.008). The differences in AUC for periapical and bitewing radiographs were 0.031 and -0.009, respectively, and were not statistically significant. Conclusions: This study found no statistically significant improvement in experienced dentists' diagnostic accuracy for detecting bone loss in intraoral radiographs when using Denti.AI deep learning technology.

Clinical validation of an artificial intelligence algorithm for classifying tuberculosis and pulmonary findings in chest radiographs

Background

Chest X-ray (CXR) interpretation is critical in diagnosing various lung diseases. However, physicians, not specialists, are often the first ones to read them, frequently facing challenges in accurate interpretation. Artificial Intelligence (AI) algorithms could be of great help, but using real-world data is crucial to ensure their effectiveness in diverse healthcare settings. This study evaluates a deep learning algorithm designed for CXR interpretation, focusing on its utility for non-specialists in thoracic radiology physicians.

Purpose

To assess the performance of a Convolutional Neural Networks (CNNs)-based AI algorithm in interpreting CXRs and compare it with a team of physicians, including thoracic radiologists, who served as the gold-standard.

Methods

A retrospective study from January 2021 to July 2023 evaluated an algorithm with three independent models for Lung Abnormality, Radiological Findings, and Tuberculosis. The algorithm's performance was measured using accuracy, sensitivity, and specificity. Two groups of physicians validated the model: one with varying specialties and experience levels in interpreting chest radiographs (Group A) and another of board-certified thoracic radiologists (Group B). The study also assessed the agreement between the two groups on the algorithm's heatmap and its influence on their decisions.

Results

In the internal validation, the Lung Abnormality and Tuberculosis models achieved an AUC of 0.94, while the Radiological Findings model yielded a mean AUC of 0.84. During the external validation, utilizing the ground truth generated by board-certified thoracic radiologists, the algorithm achieved better sensitivity in 6 out of 11 classes than physicians with varying experience levels. Furthermore, Group A physicians demonstrated higher agreement with the algorithm in identifying markings in specific lung regions than Group B (37.56% Group A vs. 21.75% Group B). Additionally, physicians declared that the algorithm did not influence their decisions in 93% of the cases.

Conclusion

This retrospective clinical validation study assesses an AI algorithm's effectiveness in interpreting Chest X-rays (CXR). The results show the algorithm's performance is comparable to Group A physicians, using gold-standard analysis (Group B) as the reference. Notably, both Groups reported minimal influence of the algorithm on their decisions in most cases.

The Future of Artificial Intelligence Using Images and Clinical Assessment for Difficult Airway Management

Artificial intelligence (AI) algorithms, particularly deep learning, are automatic and sophisticated methods that recognize complex patterns in imaging data providing high qualitative assessments. Several machine-learning and deep-learning models using imaging techniques have been recently developed and validated to predict difficult airways. Despite advances in AI modeling. In this review article, we describe the advantages of using AI models. We explore how these methods could impact clinical practice. Finally, we discuss predictive modeling for difficult laryngoscopy using machine-learning and the future approach with intelligent intubation devices.

Fine-Tuned Large Language Model for Extracting Patients on Pretreatment for Lung Cancer from a Picture Archiving and Communication System Based on Radiological Reports

This study aimed to investigate the performance of a fine-tuned large language model (LLM) in extracting patients on pretreatment for lung cancer from picture archiving and communication systems (PACS) and comparing it with that of radiologists. Patients whose radiological reports contained the term lung cancer (3111 for training, 124 for validation, and 288 for test) were included in this retrospective study. Based on clinical indication and diagnosis sections of the radiological report (used as input data), they were classified into four groups (used as reference data): group 0 (no lung cancer), group 1 (pretreatment lung cancer present), group 2 (after treatment for lung cancer), and group 3 (planning radiation therapy). Using the training and validation datasets, fine-tuning of the pretrained LLM was conducted ten times. Due to group imbalance, group 2 data were undersampled in the training. The performance of the best-performing model in the validation dataset was assessed in the independent test dataset. For testing purposes, two other radiologists (readers 1 and 2) were also involved in classifying radiological reports. The overall accuracy of the fine-tuned LLM, reader 1, and reader 2 was 0.983, 0.969, and 0.969, respectively. The sensitivity for differentiating group 0/1/2/3 by LLM, reader 1, and reader 2 was 1.000/0.948/0.991/1.000, 0.750/0.879/0.996/1.000, and 1.000/0.931/0.978/1.000, respectively. The time required for classification by LLM, reader 1, and reader 2 was 46s/2539s/1538s, respectively. Fine-tuned LLM effectively extracted patients on pretreatment for lung cancer from PACS with comparable performance to radiologists in a shorter time.

Fully and Weakly Supervised Deep Learning for Meniscal Injury Classification, and Location Based on MRI

Meniscal injury is a common cause of knee joint pain and a precursor to knee osteoarthritis (KOA). The purpose of this study is to develop an automatic pipeline for meniscal injury classification and localization using fully and weakly supervised networks based on MRI images. In this retrospective study, data were from the osteoarthritis initiative (OAI). The MR images were reconstructed using a sagittal intermediate-weighted fat-suppressed turbo spin-echo sequence. (1) We used 130 knees from the OAI to develop the LGSA-UNet model which fuses the features of adjacent slices and adjusts the blocks in Siam to enable the central slice to obtain rich contextual information. (2) One thousand seven hundred and fifty-six knees from the OAI were included to establish segmentation and classification models. The segmentation model achieved a DICE coefficient ranging from 0.84 to 0.93. The AUC values ranged from 0.85 to 0.95 in the binary models. The accuracy for the three types of menisci (normal, tear, and maceration) ranged from 0.60 to 0.88. Furthermore, 206 knees from the orthopedic hospital were used as an external validation data set to evaluate the performance of the model. The segmentation and classification models still performed well on the external validation set. To compare the diagnostic performances between the deep learning (DL) models and radiologists, the external validation sets were sent to two radiologists. The binary classification model outperformed the diagnostic performance of the junior radiologist (0.82-0.87 versus 0.74-0.88). This study highlights the potential of DL in knee meniscus segmentation and injury classification which can help improve diagnostic efficiency.

Generalizability of lesion detection and segmentation when ScaleNAS is trained on a large multi-organ dataset and validated in the liver

BACKGROUND

Tumor assessment through imaging is crucial for diagnosing and treating cancer. Lesions in the liver, a common site for metastatic disease, are particularly challenging to accurately detect and segment. This labor-intensive task is subject to individual variation, which drives interest in automation using artificial intelligence (AI).

PURPOSE

Evaluate AI for lesion detection and lesion segmentation using CT in the context of human performance on the same task. Use internal testing to determine how an AI-developed model (ScaleNAS) trained on lesions in multiple organs performs when tested specifically on liver lesions in a dataset integrating real-world and clinical trial data. Use external testing to evaluate whether ScaleNAS's performance generalizes to publicly available colorectal liver metastases (CRLM) from The Cancer Imaging Archive (TCIA).

METHODS

The CUPA study dataset included patients whose CT scan of chest, abdomen, or pelvis at Columbia University between 2010-2020 indicated solid tumors (CUIMC, n = 5011) and from two clinical trials in metastatic colorectal cancer, PRIME (n = 1183) and Amgen (n = 463). Inclusion required ≥1 measurable lesion; exclusion criteria eliminated 1566 patients. Data were divided at the patient level into training (n = 3996), validation (n = 570), and testing (n = 1529) sets. To create the reference standard for training and validation, each case was annotated by one of six radiologists, randomly assigned, who marked the CUPA lesions without access to any previous annotations. For internal testing we refined the CUPA test set to contain only patients who had liver lesions (n = 525) and formed an enhanced reference standard through expert consensus reviewing prior annotations. For external testing, TCIA-CRLM (n = 197) formed the test set. The reference standard for TCIA-CRLM was formed by consensus review of the original annotation and contours by two new radiologists. Metrics for lesion detection were sensitivity and false positives. Lesion segmentation was assessed with median Dice coefficient, under-segmentation ratio (USR), and over-segmentation ratio (OSR). Subgroup analysis examined the influence of lesion size ≥ 10  mm (measurable by RECIST1.1) versus all lesions (important for early identification of disease progression).

RESULTS

ScaleNAS trained on all lesions achieved sensitivity of 71.4% and Dice of 70.2% for liver lesions in the CUPA internal test set (3,495 lesions) and sensitivity of 68.2% and Dice 64.2% in the TCIA-CRLM external test set (638 lesions). Human radiologists had mean sensitivity of 53.5% and Dice of 73.9% in CUPA and sensitivity of 84.1% and Dice of 88.4% in TCIA-CRLM. Performance improved for ScaleNAS and radiologists in the subgroup of lesions that excluded sub-centimeter lesions.

CONCLUSIONS

Our study presents the first evaluation of ScaleNAS in medical imaging, demonstrating its liver lesion detection and segmentation performance across diverse datasets. Using consensus reference standards from multiple radiologists, we addressed inter-observer variability and contributed to consistency in lesion annotation. While ScaleNAS does not surpass radiologists in performance, it offers fast and reliable results with potential utility in providing initial contours for radiologists. Future work will extend this model to lung and lymph node lesions, ultimately aiming to enhance clinical applications by generalizing detection and segmentation across tissue types.

Combining Low-energy Images in Dual-energy Spectral CT With Deep Learning Image Reconstruction Algorithm to Improve Inferior Vena Cava Image Quality

OBJECTIVE

To explore the application of low-energy image in dual-energy spectral CT (DEsCT) combined with deep learning image reconstruction (DLIR) to improve inferior vena cava imaging.

MATERIALS AND METHODS

Thirty patients with inferior vena cava syndrome underwent contrast-enhanced upper abdominal CT with routine dose, and the 40, 50, 60, 70, and 80 keV images in the delayed phase were first reconstructed with the ASiR-V40% algorithm. Image quality was evaluated both quantitatively [CT value, SD, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR) for inferior vena cava] and qualitatively to select an optimal energy level with the best image quality. Then, the optimal-energy images were reconstructed again using deep learning image reconstruction medium strength (DLIR-M) and DLIR-H (high strength) algorithms and compared with that of ASiR-V40%.

RESULTS

The objective CT value, SD, SNR, and CNR increased with the decrease in energy level, with statistically significant differences (all P<0.05). The 40 keV images had the highest CT values, SNR, and CNR and good diagnostic acceptability, and 40 keV was selected as the best energy level. Compared with ASiR-V40% and DLIR-M, DLIR-H had the lowest SD, highest SNR and CNR, and subjective score (all P<0.001) with good consistencies between the 2 physicians (all k ≥0.75). The 40 keV images with DLIR-H had the highest overall image quality, showing sharper edges of inferior vena cava vessels and clearer lumen in patients with Budd-Chiari syndrome.

CONCLUSIONS

Compared with the ASiR-V algorithm, DLIR-H significantly reduces image noise and provides the highest CNR and best diagnostic image quality for the 40 keV DEsCT images in imaging inferior vena cava.

All You Need to Know About TACE: A Comprehensive Review of Indications, Techniques, Efficacy, Limits, and Technical Advancement

Transcatheter arterial chemoembolization (TACE) is a proven and widely accepted treatment option for hepatocellular carcinoma and it is recommended as first-line non-curative therapy for BCLC B/intermediate HCC (preserved liver function, multifocal, no cancer-related symptoms) in patients without vascular involvement. Different types of TACE are available nowadays, including TAE, c-TACE, DEB-TACE, and DSM-TACE, but at present there is insufficient evidence to recommend one TACE technique over another and the choice is left to the operator. This review then aims to provide a comprehensive overview of the current literature on indications, types of procedures, safety, and efficacy of different TACE treatments.

Machine learning applications in healthcare clinical practice and research

Machine learning (ML) is a type of artificial intelligence that assists computers in the acquisition of knowledge through data analysis, thus creating machines that can complete tasks otherwise requiring human intelligence. Among its various applications, it has proven groundbreaking in healthcare as well, both in clinical practice and research. In this editorial, we succinctly introduce ML applications and present a study, featured in the latest issue of the World Journal of Clinical Cases. The authors of this study conducted an analysis using both multiple linear regression (MLR) and ML methods to investigate the significant factors that may impact the estimated glomerular filtration rate in healthy women with and without non-alcoholic fatty liver disease (NAFLD). Their results implicated age as the most important determining factor in both groups, followed by lactic dehydrogenase, uric acid, forced expiratory volume in one second, and albumin. In addition, for the NAFLD- group, the 5th and 6th most important impact factors were thyroid-stimulating hormone and systolic blood pressure, as compared to plasma calcium and body fat for the NAFLD+ group. However, the study's distinctive contribution lies in its adoption of ML methodologies, showcasing their superiority over traditional statistical approaches (herein MLR), thereby highlighting the potential of ML to represent an invaluable advanced adjunct tool in clinical practice and research.

Evaluation of temporomandibular joint disc displacement with MRI-based radiomics analysis

OBJECTIVES

The purpose of this study was to propose a machine learning model and assess its ability to classify temporomandibular joint (TMJ) disc displacements on MR T1-weighted and proton density-weighted images.

METHODS

This retrospective cohort study included 180 TMJs from 90 patients with TMJ signs and symptoms. A radiomics platform was used to extract imaging features of disc displacements. Thereafter, different machine learning algorithms and logistic regression were implemented on radiomics features for feature selection, classification, and prediction. The radiomics features included first-order statistics, size- and shape-based features, and texture features. Six classifiers, including logistic regression, random forest, decision tree, k-nearest neighbours (KNN), XGBoost, and support vector machine were used for a model building which could predict the TMJ disc displacements. The performance of models was evaluated by sensitivity, specificity, and ROC curve.

RESULTS

KNN classifier was found to be the most optimal machine learning model for prediction of TMJ disc displacements. The AUC, sensitivity, and specificity for the training set were 0.944, 0.771, 0.918 for normal, anterior disc displacement with reduction (ADDwR) and anterior disc displacement without reduction (ADDwoR) while testing set were 0.913, 0.716, and 1 for normal, ADDwR, and ADDwoR. For TMJ disc displacements, skewness, root mean squared, kurtosis, minimum, large area low grey level emphasis, grey level non-uniformity, and long-run high grey level emphasis, were selected as optimal features.

CONCLUSIONS

This study has proposed a machine learning model by KNN analysis on TMJ MR images, which can be used for TMJ disc displacements.

AI-powered innovations in pancreatitis imaging: a comprehensive literature synthesis

Early identification of pancreatitis remains a significant clinical diagnostic challenge that impacts patient outcomes. The evolution of quantitative imaging followed by deep learning models has shown great promise in the non-invasive diagnosis of pancreatitis and its complications. We provide an overview of advancements in diagnostic imaging and quantitative imaging methods along with the evolution of artificial intelligence (AI). In this article, we review the current and future states of methodology and limitations of AI in improving clinical support in the context of early detection and management of pancreatitis.

Periodontitis diagnosis: A review of current and future trends in artificial intelligence

BACKGROUND

Artificial intelligence (AI) acts as the state-of-the-art in periodontitis diagnosis in dentistry. Current diagnostic challenges include errors due to a lack of experienced dentists, limited time for radiograph analysis, and mandatory reporting, impacting care quality, cost, and efficiency.

OBJECTIVE

This review aims to evaluate the current and future trends in AI for diagnosing periodontitis.

METHODS

A thorough literature review was conducted following PRISMA guidelines. We searched databases including PubMed, Scopus, Wiley Online Library, and ScienceDirect for studies published between January 2018 and December 2023. Keywords used in the search included "artificial intelligence," "panoramic radiograph," "periodontitis," "periodontal disease," and "diagnosis."

RESULTS

The review included 12 studies from an initial 211 records. These studies used advanced models, particularly convolutional neural networks (CNNs), demonstrating accuracy rates for periodontal bone loss detection ranging from 0.76 to 0.98. Methodologies included deep learning hybrid methods, automated identification systems, and machine learning classifiers, enhancing diagnostic precision and efficiency.

CONCLUSIONS

Integrating AI innovations in periodontitis diagnosis enhances diagnostic accuracy and efficiency, providing a robust alternative to conventional methods. These technologies offer quicker, less labor-intensive, and more precise alternatives to classical approaches. Future research should focus on improving AI model reliability and generalizability to ensure widespread clinical adoption.

Classification of Interventional Radiology Reports into Technique Categories with a Fine-Tuned Large Language Model

The aim of this study is to develop a fine-tuned large language model that classifies interventional radiology reports into technique categories and to compare its performance with readers. This retrospective study included 3198 patients (1758 males and 1440 females; age, 62.8 ± 16.8 years) who underwent interventional radiology from January 2018 to July 2024. Training, validation, and test datasets involved 2292, 250, and 656 patients, respectively. Input data involved texts in clinical indication, imaging diagnosis, and image-finding sections of interventional radiology reports. Manually classified technique categories (15 categories in total) were utilized as reference data. Fine-tuning of the Bidirectional Encoder Representations model was performed using training and validation datasets. This process was repeated 15 times due to the randomness of the learning process. The best-performed model, which showed the highest accuracy among 15 trials, was selected to further evaluate its performance in the independent test dataset. The report classification involved one radiologist (reader 1) and two radiology residents (readers 2 and 3). The accuracy and macrosensitivity (average of each category's sensitivity) of the best-performed model in the validation dataset were 0.996 and 0.994, respectively. For the test dataset, the accuracy/macrosensitivity were 0.988/0.980, 0.986/0.977, 0.989/0.979, and 0.988/0.980 in the best model, reader 1, reader 2, and reader 3, respectively. The model required 0.178 s required for classification per patient, which was 17.5-19.9 times faster than readers. In conclusion, fine-tuned large language model classified interventional radiology reports into technique categories with high accuracy similar to readers within a remarkably shorter time.

Accelerated T2W Imaging with Deep Learning Reconstruction in Staging Rectal Cancer: A Preliminary Study

Deep learning reconstruction (DLR) has exhibited potential in saving scan time. There is limited research on the evaluation of accelerated acquisition with DLR in staging rectal cancers. Our first objective was to explore the best DLR level in saving time through phantom experiments. Resolution and number of excitations (NEX) adjusted for different scan time, image quality of conventionally reconstructed T2W images were measured and compared with images reconstructed with different DLR level. The second objective was to explore the feasibility of accelerated T2W imaging with DLR in image quality and diagnostic performance for rectal cancer patients. 52 patients were prospectively enrolled to undergo accelerated acquisition reconstructed with highly-denoised DLR (DLR_H40sec) and conventional reconstruction (ConR2min). The image quality and diagnostic performance were evaluated by observers with varying experience and compared between protocols using κ statistics and area under the receiver operating characteristic curve (AUC). The phantom experiments demonstrated that DLR_H could achieve superior signal-to-noise ratio (SNR), detail conspicuity, sharpness, and less distortion within the least scan time. The DLR_H40sec images exhibited higher sharpness and SNR than ConR2min. The agreements with pathological TN-stages were improved using DLR_H40sec images compared to ConR2min (T: 0.846vs. 0.771, 0.825vs. 0.700, and 0.697vs. 0.512; N: 0.527vs. 0.521, 0.421vs. 0.348 and 0.517vs. 0.363 for junior, intermediate, and senior observes, respectively). Comparable AUCs to identify T3-4 and N1-2 tumors were achieved using DLR_H40sec and ConR2min images (P > 0.05). Consequently, with 2/3-time reduction, DLR_H40sec images showed improved image quality and comparable TN-staging performance to conventional T2W imaging for rectal cancer patients.

Super-resolution Deep Learning Reconstruction for 3D Brain MR Imaging: Improvement of Cranial Nerve Depiction and Interobserver Agreement in Evaluations of Neurovascular Conflict

RATIONALE AND OBJECTIVES

To determine if super-resolution deep learning reconstruction (SR-DLR) improves the depiction of cranial nerves and interobserver agreement when assessing neurovascular conflict in 3D fast asymmetric spin echo (3D FASE) brain MR images, as compared to deep learning reconstruction (DLR).

MATERIALS AND METHODS

This retrospective study involved reconstructing 3D FASE MR images of the brain for 37 patients using SR-DLR and DLR. Three blinded readers conducted qualitative image analyses, evaluating the degree of neurovascular conflict, structure depiction, sharpness, noise, and diagnostic acceptability. Quantitative analyses included measuring edge rise distance (ERD), edge rise slope (ERS), and full width at half maximum (FWHM) using the signal intensity profile along a linear region of interest across the center of the basilar artery.

RESULTS

Interobserver agreement on the degree of neurovascular conflict of the facial nerve was generally higher with SR-DLR (0.429-0.923) compared to DLR (0.175-0.689). SR-DLR exhibited increased subjective image noise compared to DLR (p ≥ 0.008). However, all three readers found SR-DLR significantly superior in terms of sharpness (p < 0.001); cranial nerve depiction, particularly of facial and acoustic nerves, as well as the osseous spiral lamina (p < 0.001); and diagnostic acceptability (p ≤ 0.002). The FWHM (mm)/ERD (mm)/ERS (mm-1) for SR-DLR and DLR was 3.1-4.3/0.9-1.1/8795.5-10,703.5 and 3.3-4.8/1.4-2.1/5157.9-7705.8, respectively, with SR-DLR's image sharpness being significantly superior (p ≤ 0.001).

CONCLUSION

SR-DLR enhances image sharpness, leading to improved cranial nerve depiction and a tendency for greater interobserver agreement regarding facial nerve neurovascular conflict.

Iterative Motion Correction Technique with Deep Learning Reconstruction for Brain MRI: A Volunteer and Patient Study

The aim of this study was to investigate the effect of iterative motion correction (IMC) on reducing artifacts in brain magnetic resonance imaging (MRI) with deep learning reconstruction (DLR). The study included 10 volunteers (between September 2023 and December 2023) and 30 patients (between June 2022 and July 2022) for quantitative and qualitative analyses, respectively. Volunteers were instructed to remain still during the first MRI with fluid-attenuated inversion recovery sequence (FLAIR) and to move during the second scan. IMCoff DLR images were reconstructed from the raw data of the former acquisition; IMCon and IMCoff DLR images were reconstructed from the latter acquisition. After registration of the motion images, the structural similarity index measure (SSIM) was calculated using motionless images as reference. For qualitative analyses, IMCon and IMCoff FLAIR DLR images of the patients were reconstructed and evaluated by three blinded readers in terms of motion artifacts, noise, and overall quality. SSIM for IMCon images was 0.952, higher than that for IMCoff images (0.949) (p < 0.001). In qualitative analyses, although noise in IMCon images was rated as increased by two of the three readers (both p < 0.001), all readers agreed that motion artifacts and overall quality were significantly better in IMCon images than in IMCoff images (all p < 0.001). In conclusion, IMC reduced motion artifacts in brain FLAIR DLR images while maintaining similarity to motionless images.

Using machine learning to predict carotid artery symptoms from CT angiography: A radiomics and deep learning approach

Purpose

To assess radiomics and deep learning (DL) methods in identifying symptomatic Carotid Artery Disease (CAD) from carotid CT angiography (CTA) images. We further compare the performance of these novel methods to the conventional calcium score.

Methods

Carotid CT angiography (CTA) images from symptomatic patients (ischaemic stroke/transient ischaemic attack within the last 3 months) and asymptomatic patients were analysed. Carotid arteries were classified into culprit, non-culprit and asymptomatic. The calcium score was assessed using the Agatston method. 93 radiomic features were extracted from regions-of-interest drawn on 14 consecutive CTA slices. For DL, convolutional neural networks (CNNs) with and without transfer learning were trained directly on CTA slices. Predictive performance was assessed over 5-fold cross validated AUC scores. SHAP and GRAD-CAM algorithms were used for explainability.

Results

132 carotid arteries were analysed (41 culprit, 41 non-culprit, and 50 asymptomatic). For asymptomatic vs symptomatic arteries, radiomics attained a mean AUC of 0.96(± 0.02), followed by DL 0.86(± 0.06) and then calcium 0.79(± 0.08). For culprit vs non-culprit arteries, radiomics achieved a mean AUC of 0.75(± 0.09), followed by DL 0.67(± 0.10) and then calcium 0.60(± 0.02). For multi-class classification, the mean AUCs were 0.95(± 0.07), 0.79(± 0.05), and 0.71(± 0.07) for radiomics, DL and calcium, respectively. Explainability revealed consistent patterns in the most important radiomic features.

Conclusions

Our study highlights the potential of novel image analysis techniques in extracting quantitative information beyond calcification in the identification of CAD. Though further work is required, the transition of these novel techniques into clinical practice may eventually facilitate better stroke risk stratification.

Artificial intelligence applications for immunology laboratory: image analysis and classification study of IIF photos

Artificial intelligence (AI) is increasingly being used in medicine to enhance the speed and accuracy of disease diagnosis and treatment. AI-based image analysis is expected to play a crucial role in future healthcare facilities and laboratories, offering improved precision and cost-effectiveness. As technology advances, the requirement for specialized software knowledge to utilize AI applications is diminishing. Our study will examine the advantages and challenges of employing AI-based image analysis in the field of immunology and will investigate whether physicians without software expertise can use MS Azure Portal for ANA IIF test classification and image analysis. This is the first study to perform Hep-2 image analysis using MS Azure Portal. We will also assess the potential for AI applications to aid physicians in interpreting ANA IIF results in immunology laboratories. The study was designed in four stages by two specialists. Stage 1: creation of an image library, Stage 2: finding an artificial intelligence application, Stage 3: uploading images and training artificial intelligence, Stage 4: performance analysis of the artificial intelligence application. In the first training, the average pattern identification accuracy for 72 testing images was 81.94%. After the second training, this accuracy increased to 87.5%. Patterns Precision improved from 71.42 to 79.96% after the second training. As a result, the number of correctly identified patterns and their accuracy increased with the second training process. Artificial intelligence-based image analysis shows promising potential. This technology is expected to become essential in healthcare facility laboratories, offering higher accuracy rates and lower costs.

Bridging the Gap Between Imaging and Molecular Characterization: Current Understanding of Radiomics and Radiogenomics in Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is the sixth most common malignancy worldwide and the third leading cause of cancer-related deaths. Imaging plays a crucial role in the screening, diagnosis, and monitoring of HCC; however, the potential mechanism regarding phenotypes or molecular subtyping remains underexplored. Radiomics significantly expands the selection of features available by extracting quantitative features from imaging data. Radiogenomics bridges the gap between imaging and genetic/transcriptomic information by associating imaging features with critical genes and pathways, thereby providing biological annotations to these features. Despite challenges in interpreting these connections, assessing their universality, and considering the diversity in HCC etiology and genetic information across different populations, radiomics and radiogenomics offer new perspectives for precision treatment in HCC. This article provides an up-to-date summary of the advancements in radiomics and radiogenomics throughout the HCC care continuum, focusing on the clinical applications, advantages, and limitations of current techniques and offering prospects. Future research should aim to overcome these challenges to improve the prognosis of HCC patients and leverage imaging information for patient benefit.

Leveraging 3D Convolutional Neural Networks for Accurate Recognition and Localization of Ankle Fractures

Background

Ankle fractures are common injuries with substantial implications for patient mobility and quality of life. Traditional imaging methods, while standard, have limitations in detecting subtle fractures and distinguishing them from complex bone structures. The advent of 3D Convolutional Neural Networks (3D-CNNs) offers a promising avenue for enhancing the accuracy and reliability of ankle fracture diagnoses.

Methods

In this study, we acquired 1453 high-resolution CT scans and processed them through three distinct 3D-CNN models: 3D-Mobilenet, 3D-Resnet101, and 3D-EfficientNetB7. Our approach involved meticulous preprocessing of images, including normalization and resampling, followed by a systematic comparative evaluation of the models based on accuracy, Area Under the Curve (AUC), and recall metrics. Additionally, the integration of Gradient-weighted Class Activation Mapping (Grad-CAM) provided visual interpretability of the models' predictive focus points.

Results

The 3D-EfficientNetB7 model outperformed the other models, achieving an accuracy of 0.91 and an AUC of 0.94 after 20 training epochs. It demonstrated particularly effective in the accurate detection and localization of subtle and complex fractures. Grad-CAM visualizations confirmed the model's focus on clinically relevant areas, aligning with expert assessments and enhancing trust in automated diagnostics. Spatial localization techniques were pivotal in improving interpretability, offering clear visual guidance for pinpointing fracture sites.

Conclusion

Our findings highlight the effectiveness of the 3D-EfficientNetB7 model in diagnosing ankle fractures, supported by robust performance metrics and enhanced visualization tools.

A strategy for achieving optimisation of radiological protection in digital radiology proposed by ICRP

Radiology is now predominantly a digital medium and this has extended the flexibility, efficiency and application of medical imaging. Achieving the full benefit of digital radiology requires images to be of sufficient quality to make a reliable diagnosis for each patient, while minimising risks from radiation exposure, and so involves a careful balance between competing objectives. When an optimisation programme is undertaken, a knowledge of patient doses from surveys can be valuable in identifying areas needing attention. However, any dose reduction measures must not degrade image quality to the extent that it is inadequate for the clinical purpose. The move to digital imaging has enabled versatile image acquisition and presentation, including multi-modality display and quantitative assessment, with post-processing options that adjust for optimal viewing. This means that the appearance of an image is unlikely to give any indication when the dose is higher than necessary. Moreover, options to improve performance of imaging equipment add to its complexity, so operators require extensive training to be able to achieve this. Optimisation is a continuous rather than single stage process that requires regular monitoring, review, and analysis of performance feeding into improvement and development of imaging protocols. The ICRP is in the process of publishing two reports about optimisation in digital radiology. The first report sets out components needed to ensure that a radiology service can carry optimisation through. It describes how imaging professionals should work together as a team and explains the benefits of having appropriate methodologies to monitor performance, together with the knowledge and expertise required to use them effectively. It emphasises the need for development of organisational processes that ensure tasks are carried out. The second ICRP report deals with practical requirements for optimisation of different digital radiology modalities, and builds on information provided in earlier modality specific ICRP publications.

Artificial Intelligence Efficacy as a Function of Trainee Interpreter Proficiency: Lessons from a Randomized Controlled Trial

BACKGROUND AND PURPOSE

Recently, artificial intelligence tools have been deployed with increasing speed in educational and clinical settings. However, the use of artificial intelligence by trainees across different levels of experience has not been well-studied. This study investigates the impact of artificial intelligence assistance on the diagnostic accuracy for intracranial hemorrhage and large-vessel occlusion by medical students and resident trainees.

MATERIALS AND METHODS

This prospective study was conducted between March 2023 and October 2023. Medical students and resident trainees were asked to identify intracranial hemorrhage and large-vessel occlusion in 100 noncontrast head CTs and 100 head CTAs, respectively. One group received diagnostic aid simulating artificial intelligence for intracranial hemorrhage only (n = 26); the other, for large-vessel occlusion only (n = 28). Primary outcomes included accuracy, sensitivity, and specificity for intracranial hemorrhage/large-vessel occlusion detection without and with aid. Study interpretation time was a secondary outcome. Individual responses were pooled and analyzed with the t test; differences in continuous variables were assessed with ANOVA.

RESULTS

Forty-eight participants completed the study, generating 10,779 intracranial hemorrhage or large-vessel occlusion interpretations. With diagnostic aid, medical student accuracy improved 11.0 points (P < .001) and resident trainee accuracy showed no significant change. Intracranial hemorrhage interpretation time increased with diagnostic aid for both groups (P < .001), while large-vessel occlusion interpretation time decreased for medical students (P < .001). Despite worse performance in the detection of the smallest-versus-largest hemorrhages at baseline, medical students were not more likely to accept a true-positive artificial intelligence result for these more difficult tasks. Both groups were considerably less accurate when disagreeing with the artificial intelligence or when supplied with an incorrect artificial intelligence result.

CONCLUSIONS

This study demonstrated greater improvement in diagnostic accuracy with artificial intelligence for medical students compared with resident trainees. However, medical students were less likely than resident trainees to overrule incorrect artificial intelligence interpretations and were less accurate, even with diagnostic aid, than the artificial intelligence was by itself.

The Impact of an Artificial Intelligence Certificate Program on Radiology Resident Education

RATIONALE AND OBJECTIVES

The objective of this study was to evaluate the effectiveness of a pilot artificial intelligence (AI) certificate program in aiding radiology trainees to develop an understanding of the evolving role and application of artificial intelligence in radiology. A secondary objective was set to determine the background of residents that would most benefit from such training.

MATERIALS AND METHODS

This was a prospective pilot study involving 42 radiology residents at two separate residency programs who participated in the Radiological Society of North America Imaging AI Foundational Certificate course over a four-month period. The course consisted of 6 online modules that contained didactic lectures followed by end-of-module quizzes to assess knowledge gained from these lectures. Pre- and post-course assessments were conducted to evaluate the residents' knowledge and skills in AI. Additionally, a post-course survey was performed to assess participants' overall satisfaction with the course.

RESULTS

All participating residents completed the certificate program. The mean pre-course assessment score was 37 %, which increased to 73 % after completing the modules (p < 0.001). 74 % (31/42) endorsed the belief the course improved familiarity with artificial intelligence in radiology. Residency program, residency year, and reported prior familiarity with AI were not found to influence pre-course score, post-course score, nor score improvement. 57 % (24/42) endorsed interest in pursuing further certification in AI.

CONCLUSION

Our pilot study suggests that a certificate course can effectively enhance the knowledge and skills of radiology residents in the application of AI in radiology. The benefits of such a course can be found regardless of program, resident year, and self-reported prior resident understanding of radiology in AI.

Deep-learning classification of teat-end conditions in Holstein cattle

As a means of preventing mastitis, deep learning for classifying teat-end conditions in dairy cows has not yet been optimized. By using 1426 digital images of dairy cow udders, the extent of teat-end hyperkeratosis was assessed using a four-point scale. Several deep-learning networks based on the transfer learning approach have been used to evaluate the conditions of the teat ends displayed in the digital images. The images of the teat ends were partitioned into training (70 %) and validation datasets (15 %); afterwards, the network was evaluated based on the remaining test dataset (15 %). The results demonstrated that eight different ImageNet models consistently achieved high accuracy (80.3-86.6 %). The areas under the receiver operating characteristic curves for the normal, smooth, rough, and very rough classification scores in the test data set ranged from 0.825 to 0.999. Thus, improved accuracy in image-based classification of teat tissue conditions in dairy cattle using deep learning requires more training images. This method could help farmers reduce the risks of intramammary infections, decrease the use of antimicrobials, and better manage costs associated with mastitis detection and treatment.

Contributing to the prediction of prognosis for treated hepatocellular carcinoma: Imaging aspects that sculpt the future

A novel nomogram model to predict the prognosis of hepatocellular carcinoma (HCC) treated with radiofrequency ablation and transarterial chemoembolization was recently published in the World Journal of Gastrointestinal Surgery. This model includes clinical and laboratory factors, but emerging imaging aspects, particularly from magnetic resonance imaging (MRI) and radiomics, could enhance the predictive accuracy thereof. Multiparametric MRI and deep learning radiomics models significantly improve prognostic predictions for the treatment of HCC. Incorporating advanced imaging features, such as peritumoral hypointensity and radiomics scores, alongside clinical factors, can refine prognostic models, aiding in personalized treatment and better predicting outcomes. This letter underscores the importance of integrating novel imaging techniques into prognostic tools to better manage and treat HCC.

Application of deep learning reconstruction in abdominal magnetic resonance cholangiopancreatography for image quality improvement and acquisition time reduction

PURPOSE

To compare deep learning (DL)-based and conventional reconstruction through subjective and objective analysis and ascertain whether DL-based reconstruction improves the quality and acquisition speed of clinical abdominal magnetic resonance imaging (MRI).

METHODS

The 124 patients who underwent abdominal MRI between January and July 2021 were retrospectively studied. For each patient, two-dimensional axial T2-weighted single-shot fast spin-echo MRI images with or without fat saturation were reconstructed using DL-based and conventional methods. The subjective image quality scores and objective metrics, including signal-to-noise ratios (SNRs) and contrast-to-noise ratios (CNRs) of the images were analysed. An explorative analysis was performed to compare 20 patients' MRI images with site routine settings, high-resolution settings and high-speed settings. Paired t tests and Wilcoxon signed-rank tests were used for subjective and objective comparisons.

RESULTS

A total of 144 patients were evaluated (mean age, 62.2 ± 14.1 years; 83 men). The MRI images reconstructed using DL-based methods had higher SNRs and CNRs than did those reconstructed using conventional methods (all p < 0.01). The subjective scores of the images reconstructed using DL-based methods were higher than those of the images reconstructed using conventional methods (p < 0.01), with significantly lower variation (p < 0.01). Exploratory analysis revealed that the DL-based reconstructions with thin slice thickness and higher temporal resolution had the highest image quality and were associated with the shortest scan times.

CONCLUSIONS

DL-based reconstruction methods can be used to improve the quality with higher stability and accelerate the acquisition of abdominal MRI.

Multimodal deep learning radiomics model for predicting postoperative progression in solid stage I non-small cell lung cancer

PURPOSE

To explore the application value of a multimodal deep learning radiomics (MDLR) model in predicting the risk status of postoperative progression in solid stage I non-small cell lung cancer (NSCLC).

MATERIALS AND METHODS

A total of 459 patients with histologically confirmed solid stage I NSCLC who underwent surgical resection in our institution from January 2014 to September 2019 were reviewed retrospectively. At another medical center, 104 patients were reviewed as an external validation cohort according to the same criteria. A univariate analysis was conducted on the clinicopathological characteristics and subjective CT findings of the progression and non-progression groups. The clinicopathological characteristics and subjective CT findings that exhibited significant differences were used as input variables for the extreme learning machine (ELM) classifier to construct the clinical model. We used the transfer learning strategy to train the ResNet18 model, used the model to extract deep learning features from all CT images, and then used the ELM classifier to classify the deep learning features to obtain the deep learning signature (DLS). A MDLR model incorporating clinicopathological characteristics, subjective CT findings and DLS was constructed. The diagnostic efficiencies of the clinical model, DLS model and MDLR model were evaluated by the area under the curve (AUC).

RESULTS

Univariate analysis indicated that size (p = 0.004), neuron-specific enolase (NSE) (p = 0.03), carbohydrate antigen 19 - 9 (CA199) (p = 0.003), and pathological stage (p = 0.027) were significantly associated with the progression of solid stage I NSCLC after surgery. Therefore, these clinical characteristics were incorporated into the clinical model to predict the risk of progression in postoperative solid-stage NSCLC patients. A total of 294 deep learning features with nonzero coefficients were selected. The DLS in the progressive group was (0.721 ± 0.371), which was higher than that in the nonprogressive group (0.113 ± 0.350) (p < 0.001). The combination of size、NSE、CA199、pathological stage and DLS demonstrated the superior performance in differentiating postoperative progression status. The AUC of the MDLR model was 0.885 (95% confidence interval [CI]: 0.842-0.927), higher than that of the clinical model (0.675 (95% CI: 0.599-0.752)) and DLS model (0.882 (95% CI: 0.835-0.929)). The DeLong test and decision in curve analysis revealed that the MDLR model was the most predictive and clinically useful model.

CONCLUSION

MDLR model is effective in predicting the risk of postoperative progression of solid stage I NSCLC, and it is helpful for the treatment and follow-up of solid stage I NSCLC patients.

Application of artificial intelligence model in pathological staging and prognosis of clear cell renal cell carcinoma

This study aims to develop a deep learning (DL) model based on whole-slide images (WSIs) to predict the pathological stage of clear cell renal cell carcinoma (ccRCC). The histopathological images of 513 ccRCC patients were downloaded from The Cancer Genome Atlas (TCGA) database and randomly divided into training set and validation set according to the ratio of 8∶2. The CLAM algorithm was used to establish the DL model, and the stability of the model was evaluated in the external validation set. DL features were extracted from the model to construct a prognostic risk model, which was validated in an external dataset. The results showed that the DL model showed excellent prediction ability with an area under the curve (AUC) of 0.875 and an average accuracy score of 0.809, indicating that the model could reliably distinguish ccRCC patients at different stages from histopathological images. In addition, the prognostic risk model constructed by DL characteristics showed that the overall survival rate of patients in the high-risk group was significantly lower than that in the low-risk group (P = 0.003), and AUC values for predicting 1-, 3- and 5-year overall survival rates were 0.68, 0.69 and 0.69, respectively, indicating that the prediction model had high sensitivity and specificity. The results of the validation set are consistent with the above results. Therefore, DL model can accurately predict the pathological stage and prognosis of ccRCC patients, and provide certain reference value for clinical diagnosis.

Role of Artificial intelligence model in prediction of low back pain using T2 weighted MRI of Lumbar spine

Background

Low back pain (LBP), the primary cause of disability, is the most common musculoskeletal disorder globally and the primary cause of disability. Magnetic resonance imaging (MRI) studies are inconclusive and less sensitive for identifying and classifying patients with LBP. Hence, this study aimed to investigate the role of artificial intelligence (AI) models in the prediction of LBP using T2 weighted MRI image of the lumbar spine.

Methods

This was a prospective case-control study. A total of 200 MRI patients (100 cases and controls each) referred for lumbar spine and whole spine screening were included. The scans were performed using 3.0 Tesla MRI (United Imaging Healthcare). T2 weighted images of the lumbar spine were segmented to extract radiomic features. Machine learning (ML) models, such as random forest, decision tree, logistic regression, K-nearest neighbors, adaboost, and deep learning methods (DL), such as ResNet and GoogleNet, were used, and performance measures were calculated.

Results

Our study showed that Random forest and AdaBoost are the most reliable ML models for predicting LBP. Random forest showed high performance with area under curve (AUC) values from 0.83 to 0.88 across all lumbar vertebrae and L2-L3, L3-L4, and L4-L5 intervertebral discs (IVDs), with AUCs of 0.88 the highest at L5-S1 IVD (0.92). Adaboost demonstrated high performance at the L2-L5 vertebrae with AUC values of 0.82 to 0.90, with the highest AUC (0.97) at the L5-S1 IVD. Among the DL models, GoogleNet outperformed the other models at 30 epochs with an accuracy of 0.85, followed by ResNet 18 (30 epochs) with an accuracy of 0.84.

Conclusion

The study demonstrated that ML and DL models can effectively predict LBP from MRI T2 weighted image of the lumbar spine. ML and DL models could also enhance the diagnostic accuracy of LBP, potentially leading to better patient management and outcomes.

Evaluation of the clinical application value of artificial intelligence in diagnosing head and neck aneurysms

OBJECTIVE

To evaluate the performance of a semi-automated artificial intelligence (AI) software program (CerebralDoc® system) in aneurysm detection and morphological measurement.

METHODS

In this study, 354 cases of computed tomographic angiography (CTA) were retrospectively collected in our hospital. Among them, 280 cases were diagnosed with aneurysms by either digital subtraction angiography (DSA) and CTA (DSA group, n = 102), or CTA-only (non-DSA group, n = 178). The presence or absence of aneurysms, as well as their location and related morphological features determined by AI were evaluated using DSA and radiologist findings. Besides, post-processing image quality from AI and radiologists were also rated and compared.

RESULTS

In the DSA group, AI achieved a sensitivity of 88.24% and an accuracy of 81.97%, whereas radiologists achieved a sensitivity of 95.10% and an accuracy of 84.43%, using DSA results as the gold standard. The AI in the non-DSA group achieved 81.46% sensitivity and 76.29% accuracy, as per the radiologists' findings. The comparison of position consistency results showed better performance under loose criteria than strict criteria. In terms of morphological characteristics, both the DSA and the non-DSA groups agreed well with the diagnostic results for neck width and maximum diameter, demonstrating excellent ICC reliability exceeding 0.80. The AI-generated images exhibited superior quality compared to the standard software for post-processing, while also demonstrating a significantly reduced processing time.

CONCLUSIONS

The AI-based aneurysm detection rate demonstrates a commendable performance, while the extracted morphological parameters exhibit a remarkable consistency with those assessed by radiologists, thereby showcasing significant potential for clinical application.

Automated abdominal CT contrast phase detection using an interpretable and open-source artificial intelligence algorithm

OBJECTIVES

To develop and validate an open-source artificial intelligence (AI) algorithm to accurately detect contrast phases in abdominal CT scans.

MATERIALS AND METHODS

Retrospective study aimed to develop an AI algorithm trained on 739 abdominal CT exams from 2016 to 2021, from 200 unique patients, covering 1545 axial series. We performed segmentation of five key anatomic structures-aorta, portal vein, inferior vena cava, renal parenchyma, and renal pelvis-using TotalSegmentator, a deep learning-based tool for multi-organ segmentation, and a rule-based approach to extract the renal pelvis. Radiomics features were extracted from the anatomical structures for use in a gradient-boosting classifier to identify four contrast phases: non-contrast, arterial, venous, and delayed. Internal and external validation was performed using the F1 score and other classification metrics, on the external dataset "VinDr-Multiphase CT".

RESULTS

The training dataset consisted of 172 patients (mean age, 70 years ± 8, 22% women), and the internal test set included 28 patients (mean age, 68 years ± 8, 14% women). In internal validation, the classifier achieved an accuracy of 92.3%, with an average F1 score of 90.7%. During external validation, the algorithm maintained an accuracy of 90.1%, with an average F1 score of 82.6%. Shapley feature attribution analysis indicated that renal and vascular radiodensity values were the most important for phase classification.

CONCLUSION

An open-source and interpretable AI algorithm accurately detects contrast phases in abdominal CT scans, with high accuracy and F1 scores in internal and external validation, confirming its generalization capability.

CLINICAL RELEVANCE STATEMENT

Contrast phase detection in abdominal CT scans is a critical step for downstream AI applications, deploying algorithms in the clinical setting, and for quantifying imaging biomarkers, ultimately allowing for better diagnostics and increased access to diagnostic imaging.

KEY POINTS

Digital Imaging and Communications in Medicine labels are inaccurate for determining the abdominal CT scan phase. AI provides great help in accurately discriminating the contrast phase. Accurate contrast phase determination aids downstream AI applications and biomarker quantification.