Imaging Of Lesions Research Articles

A deep learning approach to detection of oral cancer lesions from intra oral patient images: A preliminary retrospective study

IntroductionOral squamous cell carcinomas (OSCC) seen in the oral cavity are a category of diseases for which dentists may diagnose and even cure. This study evaluated the performance of diagnostic computer software developed to detect oral cancer lesions in intra-oral retrospective patient images. Materials and methodsOral cancer lesions were labeled with CranioCatch labeling program (CranioCatch, Eskişehir, Turkey) and polygonal type labeling method on a total of 65 anonymous retrospective intraoral patient images of oral mucosa that were diagnosed with oral cancer histopathologically by incisional biopsy from individuals in our clinic. All images have been rechecked and verified by experienced experts. This data set was divided into training (n = 53), validation (n = 6) and test (n = 6) sets. Artificial intelligence model was developed using YOLOv5 architecture, which is a deep learning approach. Model success was evaluated with confusion matrix. ResultsWhen the success rate in estimating the images reserved for the test not used in education was evaluated, the F1, sensitivity and precision results of the artificial intelligence model obtained using the YOLOv5 architecture were found to be 0.667, 0.667 and 0.667, respectively. ConclusionsOur study reveals that OCSCC lesions carry discriminative visual appearances, which can be identified by deep learning algorithm. Artificial intelligence shows promise in the prediagnosis of oral cancer lesions. The success rates will increase in the training models of the data set that will be formed with more images.

Safety and effectiveness of disease-modifying therapies after switching from natalizumab

Introduction: One strategy to mitigate progressive multifocal leukoencephalopathy (PML) risk is to switch to other highly effective disease-modifying therapies (DMTs). However, the optimal switch DMT following natalizumab (NTZ) discontinuation is yet to be determined. Objective: The objective of the study is to determine the most effective and tolerable DMTs to switch to following NTZ discontinuation due to John Cunningham virus (JCV) antibody positivity. Methods: This is a multicenter observational cohort study that included all stable relapsing-remitting multiple sclerosis (MS) patients who were treated with NTZ for at least 6 months before switching therapy due to JCV antibody positivity. Results: Of 321 patients, 255 switched from NTZ to rituximab/ocrelizumab, 52 to fingolimod, and 14 to alemtuzumab, with higher annualized relapse rate (ARR) in fingolimod switchers (0.193) compared with rituximab/ocrelizumab or alemtuzumab (0.028 and 0.032, respectively). Fingolimod switchers also had increased disability progression (p = 0.014) and a higher proportion developed magnetic resonance imaging (MRI) lesions compared with rituximab/ocrelizumab (62.9% vs. 13.0%, p < 0.001, and 66.6% vs. 24.0%, p < 0.001, respectively). Mean drug survival favored rituximab/ocrelizumab or alemtuzumab over fingolimod (p < 0.001). Conclusion: Our study shows superior effectiveness of rituximab/ocrelizumab and alemtuzumab compared with fingolimod in stable patients switching from NTZ due to JC virus antibody positivity.

Neuroimaging and clinical features of bilateral Wallerian degeneration of middle cerebellar peduncles subsequent to pontine infarction.

Wallerian degeneration (WD) of the middle cerebellar peduncles (MCPs) following pontine infarction is a rare secondary degenerative neurological condition. Due to its infrequency, there is limited research on its characteristics. This study aims to present three cases of WD of MCPs following pontine infarction and to analyze the prognosis, clinical manifestations, and neuroimaging features by amalgamating our cases with previously reported ones. The cohort consisted of 25 cases, comprising 18 men and 7 women aged 29 to 77 years (mean age: 66.2 years). The majority of patients (94%) exhibit risk factors for cerebrovascular disease, with hypertension being the primary risk factor. Magnetic resonance imaging (MRI) can detect WD of MCPs within a range of 21 days to 12 months following pontine infarction. This degeneration is characterized by bilateral symmetric hyperintensities on T2/FLAIR-weighted images (WI) lesions in the MCPs. Moreover, restricted diffusion, with hyperintensity on diffusion-weighted imaging (DWI) and low apparent diffusion coefficient (ADC) signal intensity may be observed as early as 21 days after the infarction. Upon detection of WD, it was observed that 20 patients (80%) remained asymptomatic during subsequent clinic visits, while four (16%) experienced a worsening of pre-existing symptoms. These findings underscore the importance of neurologists enhancing their understanding of this condition by gaining fresh insights into the neuroimaging characteristics, clinical manifestations, and prognosis of individuals with WD of bilateral MCPs.

BT10 (P025) Community lesion imaging clinics for tele-triage of suspected skin cancer: implementation of a low-cost model in an NHS trust

BT10 (P025) Community lesion imaging clinics for tele-triage of suspected skin cancer: implementation of a low-cost model in an NHS trust

Distributable, metabolic PET reporting of tuberculosis

Tuberculosis remains a large global disease burden for which treatment regimens are protracted and monitoring of disease activity difficult. Existing detection methods rely almost exclusively on bacterial culture from sputum which limits sampling to organisms on the pulmonary surface. Advances in monitoring tuberculous lesions have utilized the common glucoside [18F]FDG, yet lack specificity to the causative pathogen Mycobacterium tuberculosis (Mtb) and so do not directly correlate with pathogen viability. Here we show that a close mimic that is also positron-emitting of the non-mammalian Mtb disaccharide trehalose – 2-[18F]fluoro-2-deoxytrehalose ([18F]FDT) – is a mechanism-based reporter of Mycobacteria-selective enzyme activity in vivo. Use of [18F]FDT in the imaging of Mtb in diverse models of disease, including non-human primates, successfully co-opts Mtb-mediated processing of trehalose to allow the specific imaging of TB-associated lesions and to monitor the effects of treatment. A pyrogen-free, direct enzyme-catalyzed process for its radiochemical synthesis allows the ready production of [18F]FDT from the most globally-abundant organic 18F-containing molecule, [18F]FDG. The full, pre-clinical validation of both production method and [18F]FDT now creates a new, bacterium-selective candidate for clinical evaluation. We anticipate that this distributable technology to generate clinical-grade [18F]FDT directly from the widely-available clinical reagent [18F]FDG, without need for either custom-made radioisotope generation or specialist chemical methods and/or facilities, could now usher in global, democratized access to a TB-specific PET tracer.

Multiparametric Prostate MRI Accuracy of Prostate Imaging Reporting and Data System (v2.1) Scores 4 and 5: The Influence of Image Quality According to the Prostate Imaging Quality Score.

Purpose: The accuracy of multiparametric magnetic resonance imaging (mpMRI) heavily relies on image quality, as evidenced by the evolution of the prostate imaging quality (PI-QUAL) scoring system for the evaluation of clinically significant prostate cancer (csPC). This study aims to evaluate the impact of PI-QUAL scores in detecting csPC within PI-RADS 4 and 5 lesions. Methods: We retrospectively selected from our database all mpMRI performed from January 2019 to March 2022. Inclusion criteria were as follows: (1) mpMRI acquired in our institution according to the technical requirements from the PI-RADS (v2.1) guidelines; (2) single lesion scored as PI-RADS (v2.1) 4 or 5; (3) MRI-TBx performed in our institution; (4) complete histology report; and (5) complete clinical record. Results: A total of 257 male patients, mean age 70.42 ± 7.6 years, with a single PI-RADS 4 or 5 lesion undergoing MRI-targeted biopsy, were retrospectively studied. Of these, 61.5% were PI-RADS 4, and 38.5% were PI-RADS 5, with 84% confirming neoplastic cells. In high-quality image lesions (PI-QUAL ≥ 4), all PI-RADS 5 lesions were accurately identified as positive at the final histological examination (100% of CDR). For PI-RADS 4 lesions, 37 (23%) were negative, resulting in a cancer detection rate of 77% (95% CI: 67.51-84.83). Conclusions: The accuracy of mpMRI, independently of the PI-RADS score, progressively decreased according to the decreasing PI-QUAL score. These findings emphasize the crucial role of the PI-QUAL scoring system in evaluating PI-RADS 4 and 5 lesions, influencing mpMRI accuracy.

Deep learning image reconstruction generates thinner slice iodine maps with improved image quality to increase diagnostic acceptance and lesion conspicuity: a prospective study on abdominal dual-energy CT

BackgroundTo assess the improvement of image quality and diagnostic acceptance of thinner slice iodine maps enabled by deep learning image reconstruction (DLIR) in abdominal dual-energy CT (DECT).MethodsThis study prospectively included 104 participants with 136 lesions. Four series of iodine maps were generated based on portal-venous scans of contrast-enhanced abdominal DECT: 5-mm and 1.25-mm using adaptive statistical iterative reconstruction-V (Asir-V) with 50% blending (AV-50), and 1.25-mm using DLIR with medium (DLIR-M), and high strength (DLIR-H). The iodine concentrations (IC) and their standard deviations of nine anatomical sites were measured, and the corresponding coefficient of variations (CV) were calculated. Noise-power-spectrum (NPS) and edge-rise-slope (ERS) were measured. Five radiologists rated image quality in terms of image noise, contrast, sharpness, texture, and small structure visibility, and evaluated overall diagnostic acceptability of images and lesion conspicuity.ResultsThe four reconstructions maintained the IC values unchanged in nine anatomical sites (all p > 0.999). Compared to 1.25-mm AV-50, 1.25-mm DLIR-M and DLIR-H significantly reduced CV values (all p < 0.001) and presented lower noise and noise peak (both p < 0.001). Compared to 5-mm AV-50, 1.25-mm images had higher ERS (all p < 0.001). The difference of the peak and average spatial frequency among the four reconstructions was relatively small but statistically significant (both p < 0.001). The 1.25-mm DLIR-M images were rated higher than the 5-mm and 1.25-mm AV-50 images for diagnostic acceptability and lesion conspicuity (all P < 0.001).ConclusionsDLIR may facilitate the thinner slice thickness iodine maps in abdominal DECT for improvement of image quality, diagnostic acceptability, and lesion conspicuity.

Lentivirus-modified hematopoietic stem cell gene therapy for advanced symptomatic juvenile metachromatic leukodystrophy: A long-term follow-up pilot study.

Metachromatic leukodystrophy (MLD) is an inherited disease caused by a deficiency of the enzyme arylsulfatase A (ARSA). Lentivirus-modified autologous hematopoietic stem cell gene therapy (HSCGT) has recently been approved for clinical use in pre- and early-symptomatic children with MLD to increase ARSA activity. Unfortunately, this advanced therapy is not available for most patients with MLD who have progressed to more advanced symptomatic stages at diagnosis. Patients with late-onset juvenile MLD typically present with a slower neurological progression of symptoms and represent a significant burden to the economy and healthcare system, whereas those with early-onset infantile MLD die within a few years of symptom onset. We conducted a pilot study to determine the safety and benefit of HSCGT in patients with post-symptomatic juvenile MLD and report preliminary results. The safety profile of HSCGT was favorable in this long-term follow-up over nine years. The most common adverse events (AEs) within two months of HSCGT were related to busulfan conditioning, and all AEs resolved. No HSCGT-related AEs and no evidence of distorted hematopoietic differentiation during long-term follow-up for up to 9.6 years. Importantly, to date, patients have maintained remarkably improved ARSA activity with a stable disease state, including increased Functional Independence Measure (FIM) score and decreased magnetic resonance imaging (MRI) lesion score. This long-term follow-up pilot study suggests that HSCGT is safe and provides clinical benefit to patients with post-symptomatic juvenile MLD.

Skin lesion classification with multi-level fusion of Swin-T and ConvNeXt

Skin cancer is a significant public health issue, and computer-aided diagnosis technology can effectively alleviate this burden. Accurate identification of skin lesion types is crucial when employing computer-aided diagnosis. This study proposes a multi-level attention cascaded fusion model based on Swin-T and ConvNeXt. It employed hierarchical Swin-T and ConvNeXt to extract global and local features, respectively, and introduced residual channel attention and spatial attention modules for further feature extraction. Multi-level attention mechanisms were utilized to process multi-scale global and local features. To address the problem of shallow features being lost due to their distance from the classifier, a hierarchical inverted residual fusion module was proposed to dynamically adjust the extracted feature information. Balanced sampling strategies and focal loss were employed to tackle the issue of imbalanced categories of skin lesions. Experimental testing on the ISIC2018 and ISIC2019 datasets yielded accuracy, precision, recall, and F1-Score of 96.01%, 93.67%, 92.65%, and 93.11%, respectively, and 92.79%, 91.52%, 88.90%, and 90.15%, respectively. Compared to Swin-T, the proposed method achieved an accuracy improvement of 3.60% and 1.66%, and compared to ConvNeXt, it achieved an accuracy improvement of 2.87% and 3.45%. The experiments demonstrate that the proposed method accurately classifies skin lesion images, providing a new solution for skin cancer diagnosis.

Advancing Skin Cancer Prediction Using Ensemble Models

There are many different kinds of skin cancer, and an early and precise diagnosis is crucial because skin cancer is both frequent and deadly. The key to effective treatment is accurately classifying the various skin cancers, which have unique traits. Dermoscopy and other advanced imaging techniques have enhanced early detection by providing detailed images of lesions. However, accurately interpreting these images to distinguish between benign and malignant tumors remains a difficult task. Improved predictive modeling techniques are necessary due to the frequent occurrence of erroneous and inconsistent outcomes in the present diagnostic processes. Machine learning (ML) models have become essential in the field of dermatology for the automated identification and categorization of skin cancer lesions using image data. The aim of this work is to develop improved skin cancer predictions by using ensemble models, which combine numerous machine learning approaches to maximize their combined strengths and reduce their individual shortcomings. This paper proposes a fresh and special approach for ensemble model optimization for skin cancer classification: the Max Voting method. We trained and assessed five different ensemble models using the ISIC 2018 and HAM10000 datasets: AdaBoost, CatBoost, Random Forest, Gradient Boosting, and Extra Trees. Their combined predictions enhance the overall performance with the Max Voting method. Moreover, the ensemble models were fed with feature vectors that were optimally generated from the image data by a genetic algorithm (GA). We show that, with an accuracy of 95.80%, the Max Voting approach significantly improves the predictive performance when compared to the five ensemble models individually. Obtaining the best results for F1-measure, recall, and precision, the Max Voting method turned out to be the most dependable and robust. The novel aspect of this work is that skin cancer lesions are more robustly and reliably classified using the Max Voting technique. Several pre-trained machine learning models’ benefits are combined in this approach.

The Japanese Esophageal Society classification for prediction of superficial esophageal squamous cell neoplasia invasion depth: Validation in a Western population.

The Japan Esophageal Society proposed the JES microvessel classification to assess eligibility of early esophageal squamous cell neoplasia (ESCN) for endoscopic resection based on intrapapillary capillary loop assessment. We aimed to assess its diagnostic reproducibility and accuracy in Western ESCN patients. Intrapapillary capillary loops on endoscopic images of Western ESCN lesions (n=113) collected between 2010 and 2022 were assessed by nine endoscopists, including three Japanese expert endoscopists, three Western expert endoscopists, and three residents-in-training, and graded according to the JES microvessel classification where microvessel type A corresponds with normality or low-grade intraepithelial neoplasia, and microvessel types B1, B2, and B3 correspond with high-grade intraepithelial neoplasia or invasion into the lamina propria, muscularis mucosae or superficial submucosa, and deep submucosa, respectively. Outcomes included overall accuracy in predicting ESCN invasion depth and interobserver agreement. Good interobserver agreement was observed among expert endoscopists (Krippendorf's alpha 0.64, 95% CI 0.57-0.70), while agreement was moderate among residents-in-training (Krippendorf's alpha 0.58, 95% CI 0.52-0.72). Overall accuracy of the JES microvessel classification was 53% (95% CI 42-63), 52% (95% CI 41-62), and 44% (95% CI 34-55) for Japanese endoscopists, Western endoscopists, and residents-in-training, respectively. Sensitivity and specificity for vessel type A, B1, B2, and B3 across assessors were 0%-50% and 89%-100%, 55%-64% and 66%-77%, 42%-71% and 60%-76%, and 10%-24% and 92%-97%, respectively. Negative predictive value ranged between 80% and 85% for B3 vessels. Overall accuracy of the JES microvessel classification in Western ESCN patients is low, though absence of B3 vessels as assessed by experienced endoscopists may predict superficial ESCN amenable to endoscopic resection. www.trialregister.nl; NL8897 (6-9-2020).

Detection and classification of breast lesions in thermographic images by deep transfer learning hybrid architectures

Detection and classification of breast lesions in thermographic images by deep transfer learning hybrid architectures

Intracranial cerebrovascular lesions on T2-weighted magnetic resonance imaging.

Magnetic resonance imaging (MRI) of the brain has been implemented to evaluate multiple intracranial pathologies. Non-contrast T2-weighted images are a routinely acquired sequence in almost all neuroimaging protocols. It is not uncommon to encounter various cerebrovascular lesions incidentally on brain imaging. Neuroradiologists should evaluate the routine T2-weighted images for incidental cerebrovascular lesions, irrespective of the primary indication of the study. Vascular structures typically demonstrate a low signal flow-void on the T2-weighted images. In our experience, large cerebrovascular abnormalities are easily visible to a typical neuroradiologist. In this article, we present the spectrum of the characteristic imaging appearance of various intracranial cerebrovascular lesions on routine non-contrast T2-weighted MRI. These include aneurysm, arteriovenous malformation, arterial occlusion, capillary telangiectasia, cavernous malformation, dural arteriovenous fistula, moyamoya, proliferative angiopathy, and vein of Galen malformation.

Automated interpretation of retinal vein occlusion based on fundus fluorescein angiography images using deep learning: A retrospective, multi-center study

PurposeFundus fluorescein angiography (FFA) is the gold standard for retinal vein occlusion (RVO) diagnosis. This study aims to develop a deep learning-based system to diagnose and classify RVO using FFA images, addressing the challenges of time-consuming and variable interpretations by ophthalmologists. Methods4028 FFA images of 467 eyes from 463 patients were collected and annotated. Three convolutional neural networks (CNN) models (ResNet50, VGG19, InceptionV3) were trained to generate the label of image quality, eye, location, phase, lesions, diagnosis, and macular involvement. The performance of the models was evaluated by accuracy, precision, recall, F-1 score, the area under the curve, confusion matrix, human-machine comparison, and Clinical validation on three external data sets. ResultsThe InceptionV3 model outperformed ResNet50 and VGG19 in labeling and interpreting FFA images for RVO diagnosis, achieving 77.63%–96.45% accuracy for basic information labels and 81.72%–96.45% for RVO-relevant labels. The comparison between the best CNN and ophthalmologists showed up to 19% accuracy improvement with the inceptionV3. ConclusionThis study developed a deep learning model capable of automatically multi-label and multi-classification of FFA images for RVO diagnosis. The proposed system is anticipated to serve as a new tool for diagnosing RVO in places short of medical resources.

Noninvasive prediction of lymph node metastasis in pancreatic cancer using an ultrasound-based clinicoradiomics machine learning model

ObjectivesThis study was designed to explore and validate the value of different machine learning models based on ultrasound image-omics features in the preoperative diagnosis of lymph node metastasis in pancreatic cancer (PC).MethodsThis research involved 189 individuals diagnosed with PC confirmed by surgical pathology (training cohort: n = 151; test cohort: n = 38), including 50 cases of lymph node metastasis. Image-omics features were extracted from ultrasound images. After dimensionality reduction and screening, eight machine learning algorithms, including logistic regression (LR), support vector machine (SVM), K-nearest neighbors (KNN), random forest (RF), extra trees (ET), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and multilayer perceptron (MLP), were used to establish image-omics models to predict lymph node metastasis in PC. The best omics prediction model was selected through ROC curve analysis. Machine learning models were used to analyze clinical features and determine variables to establish a clinical model. A combined model was constructed by combining ultrasound image-omics and clinical features. Decision curve analysis (DCA) and a nomogram were used to evaluate the clinical application value of the model.ResultsA total of 1561 image-omics features were extracted from ultrasound images. 15 valuable image-omics features were determined by regularization, dimension reduction, and algorithm selection. In the image-omics model, the LR model showed higher prediction efficiency and robustness, with an area under the ROC curve (AUC) of 0.773 in the training set and an AUC of 0.850 in the test set. The clinical model constructed by the boundary of lesions in ultrasound images and the clinical feature CA199 (AUC = 0.875). The combined model had the best prediction performance, with an AUC of 0.872 in the training set and 0.918 in the test set. The combined model showed better clinical benefit according to DCA, and the nomogram score provided clinical prediction solutions.ConclusionThe combined model established with clinical features has good diagnostic ability and can be used to predict lymph node metastasis in patients with PC. It is expected to provide an effective noninvasive method for clinical decision-making, thereby improving the diagnosis and treatment of PC.

Evaluation of Two Digital Wound Area Measurement Methods Using a Non-Randomized, Single-Center, Controlled Clinical Trial

A prospective, single-center, non-randomized, pre-marketing clinical investigation was conducted with a single group of subjects to collect skin lesion images. These images were subsequently utilized to compare the results obtained from a traditional method of wound size measurement with two novel methods developed using Machine Learning (ML) approaches. Both proposed methods automatically calculate the wound area from an image. One method employs a two-dimensional system with the assistance of an external calibrator, while the other utilizes an Augmented Reality (AR) system, eliminating the need for a physical calibration object. To validate the correlation between these methods, a gold standard measurement with digital planimetry was employed. A total of 67 wound images were obtained from 41 patients between 22 November 2022 and 10 February 2023. The conducted pre-marketing clinical investigation demonstrated that the ML algorithms are safe for both the intended user and the intended target population. They exhibit a high correlation with the gold standard method and are more accurate than traditional methods. Additionally, they meet the manufacturer’s expected use. The study validated the performance, safety, and usability of the implemented methods as a valuable tool in the measurement of skin lesions.

Contrast-enhanced ultrasound (CEUS): applications from the kidneys to the bladder.

Contrast-enhanced ultrasound (CEUS) is an advanced ultrasound (US) technique utilizing ultrasound contrast agents (UCAs) to provide detailed visualization of anatomic and vascular architecture, including the depiction of microcirculation. CEUS has been well-established in echocardiography and imaging of focal hepatic lesions and recent studies have also shown the utility of CEUS in non-hepatic applications like the urinary system. The updated guidelines by the European Federation of Societies for Ultrasound in Medicine and Biology (EFSUMB) from 2018 describe the use of CEUS for non-hepatic applications. CEUS' excellent safety profile and spatial resolution make it a superior modality to conventional US and is often comparable and even superior to CECT in some instances. In comparison to other cross-sectional imaging modalities such as CECT or MRI, CEUS offers a safe (by virtue of non-nephrotoxic US contrast agents), accurate, cost-efficient, readily available, and a quick means of evaluation of multiple pathologies of the urinary system. CEUS also has the potential to reduce the overall economic burden on patients requiring long-term follow-up due to its low cost as compared to CT or MRI techniques. This comprehensive review focuses on the applications of CEUS in evaluating the urinary system from the kidneys to the urinary bladder. CEUS can be utilized in the kidney to evaluate complex cystic lesions, indeterminate lesions, pseudotumors (vs solid renal tumors), renal infections, and renal ischemic disorders. Additionally, CEUS has also been utilized in evaluating renal transplants. In the urinary bladder, CEUS is extremely useful in differentiating a bladder hematoma and bladder cancer when conventional US techniques show equivocal results. Quantitative parameters of time-intensity curves (TICs) of CEUS examinations have also been studied to stage and grade bladder cancers. Although promising, further research is needed to definitively stage bladder cancers and classify them as muscle-invasive or non-muscle invasive using quantitative CEUS to guide appropriate intervention. CEUS has been very effective in the classification of cystic renal lesions, however, further research is needed in differentiating benign from malignant renal masses.

MRI Radiomics in Imaging of Focal Hepatic Lesions: A Narrative Review.

Magnetic resonance imaging (MRI) is generally used to identify, describe, and evaluate treatment responses for focal hepatic lesions.However, the diagnosis and differentiation of such lesions require considerable input from radiologists. In order to reduce these difficulties, radiomics is an artificial intelligence (AI)-based quantitative method that employs the extraction of image features to reliably detect and differentiate focal hepatic lesions. MRI radiomics is a novel technique for the characterization of focal hepatic lesions. It can aid in preoperative evaluation, treatment approach, and forecast microvascular invasion. Although many studies have illustrated its efficiency there are certain limitations such as the absence of a large diverse dataset, comparison with other AI models, integration with histopathological findings, clinical utility, and feasibility.

Panoramic heat map for spatial distribution of necrotic lesions.

In this study, we aimed to visualize the spatial distribution characteristics of femoral head necrosis using a novel measurement method. We retrospectively collected CT imaging data of 108 hips with non-traumatic osteonecrosis of the femoral head from 76 consecutive patients (mean age 34.3 years (SD 8.1), 56.58% male (n = 43)) in two clinical centres. The femoral head was divided into 288 standard units (based on the orientation of units within the femoral head, designated as N[Superior], S[Inferior], E[Anterior], and W[Posterior]) using a new measurement system called the longitude and latitude division system (LLDS). A computer-aided design (CAD) measurement tool was also developed to visualize the measurement of the spatial location of necrotic lesions in CT images. Two orthopaedic surgeons independently performed measurements, and the results were used to draw 2D and 3D heat maps of spatial distribution of necrotic lesions in the femoral head, and for statistical analysis. The results showed that the LLDS has high inter-rater reliability. As illustrated by the heat map, the distribution of Japanese Investigation Committee (JIC) classification type C necrotic lesions exhibited clustering characteristics, with the lesions being concentrated in the northern and eastern regions, forming a hot zone (90% probability) centred on the N4-N6E2, N3-N6E units of outer ring blocks. Statistical results showed that the distribution difference between type C2 and type C1 was most significant in the E1 and E2 units and, combined with the heat map, indicated that the spatial distribution differences at N3-N6E1 and N1-N3E2 units are crucial in understanding type C1 and C2 necrotic lesions. The LLDS can be used to accurately measure the spatial location of necrotic lesions and display their distribution characteristics.

Y-90 PET/MR imaging optimization with a Bayesian penalized likelihood reconstruction algorithm.

Positron Emission Tomography (PET) imaging after Y liver radioembolization is used for both lesion identification and dosimetry. Bayesian penalized likelihood (BPL) reconstruction algorithms are an alternative to ordered subset expectation maximization (OSEM) with improved image quality and lesion detectability. The investigation of optimal parameters for Y image reconstruction of Q.Clear, a commercial BPL algorithm developed by General Electric (GE), in PET/MR is a field of interest and the subject of this study. The NEMA phantom was filled at an 8:1 sphere-to-background ratio. Acquisitions were performed on a PET/MR scanner for clinically relevant activities between 0.7 and 3.3 MBq/ml. Reconstructions with Q.Clear were performed varying the penalty parameter between 20 and 6000, the acquisition time between 5 and 20 min and pixel size between 1.56 and 4.69 mm. OSEM reconstructions of 28 subsets with 2 and 4 iterations with and without Time-of-Flight (TOF) were compared to Q.Clear with = 4000. Recovery coefficients (RC), their coefficient of variation (COV), background variability (BV), contrast-to-noise ratio (CNR) and residual activity in the cold insert were evaluated. Increasing parameter lowered RC, COV and BV, while CNR was maximized at = 4000; further increase resulted in oversmoothing. For quantification purposes, = 1000-2000 could be more appropriate. Longer acquisition times resulted in larger CNR due to reduced image noise. Q.Clear reconstructions led to higher CNR than OSEM. A of 4000 was obtained for optimal image quality, although lower values could be considered for quantification purposes. An optimal acquisition time of 15 min was proposed considering its clinical use.