Performance Evaluation of Deep, Shallow and Ensemble Machine Learning Methods for the Automated Classification of Alzheimer's Disease.

Artificial intelligence (AI)-based approaches are crucial in computer-aided diagnosis (CAD) for various medical applications. Their ability to quickly and accurately learn from complex data is remarkable. Deep learning (DL) models have shown promising results in accurately classifying Alzheimer's disease (AD) and its related cognitive states, Early Mild Cognitive Impairment (EMCI) and Late Mild Cognitive Impairment (LMCI), along with the healthy conditions known as Cognitively Normal (CN). This offers valuable insights into disease progression and diagnosis. However, certain traditional machine learning (ML) classifiers perform equally well or even better than DL models, requiring less training data. This is particularly valuable in CAD in situations with limited labeled datasets. In this paper, we propose an ensemble classifier based on ML models for magnetic resonance imaging (MRI) data, which achieved an impressive accuracy of 96.52%. This represents a 3-5% improvement over the best individual classifier. We evaluated popular ML classifiers for AD classification under both data-scarce and data-rich conditions using the Alzheimer's Disease Neuroimaging Initiative and Open Access Series of Imaging Studies datasets. By comparing the results to state-of-the-art CNN-centric DL algorithms, we gain insights into the strengths and weaknesses of each approach. This work will help users to select the most suitable algorithm for AD classification based on data availability.

Characterization of white matter lesions in multiple sclerosis using proton density and T1-relaxation measures.

PURPOSE: Although lesion dissemination in time is a defining characteristic of multiple sclerosis (MS), there is a limited understanding of lesion heterogeneity. Currently, conventional sequences such as fluid attenuated inversion recovery (FLAIR) and T1-weighted (T1W) data are used to assess MS lesions qualitatively. Estimating water content could provide a measure of local tissue rarefaction, or reduced tissue density, resulting from chronic inflammation. Our goal was to utilize the proton spin density (PD), derived from a rapid, multi-contrast STAGE (strategically acquired gradient echo) protocol to characterize white matter (WM) lesions seen on T2W, FLAIR and T1W data. MATERIALS AND METHODS: Twenty (20) subjects with relapsing-remitting MS were scanned at 3 T using T1W, T2-weighted, FLAIR and strategically acquired gradient echo (STAGE) sequences. PD and T1 maps were derived from the STAGE data. Disease severity scores, including Extended Disability Status Scale (EDSS) and Multiple Sclerosis Functional Composite (MSFC), were correlated with total, high PD and high T1 lesion volumes. A probability map of high PD regions and all lesions across all subjects was generated. Five perilesional normal appearing WM (NAWM) bands surrounding the lesions were generated to compare the median PD and T1 values in each band with the lesional values and the global WM. RESULTS: T1W intensity was negatively correlated with PD as expected (R = -0.87, p < 0.01, R2 = 0.756) and the FLAIR signal was suppressed for high PD volumes within the lesions, roughly for PD ≥ 0.85. The threshold for high PD and T1 regions was set to 0.909 and 1953.6 ms, respectively. High PD regions showed a high probability of occurrence near the boundary of the lateral ventricles. EDSS score and nine-hole peg test (dominant and non-dominant hand) were significantly correlated with the total lesion volume and the volumes of high PD and T1 regions (p < 0.05). There was a significant difference in PD/T1 values between the high PD/T1 regions within the lesions and the remaining lesional tissue (p < 0.001). In addition, the PD values of the first NAWM perilesional band directly adjacent to the lesional boundary displayed a significant difference (p < 0.05) compared to the global WM. CONCLUSION: Lesions with high PD and T1s had the highest probability of occurrence at the boundary of the lateral ventricles and likely represent chronic lesions with significant local tissue rarefaction. Moreover, the perilesional NAWM exhibited subtly increasing PD and T1 values from the NAWM up to the lesion boundary. Unlike on the T1 maps, the perilesional band adjacent to the lesion boundary possessed a significantly higher PD value than the global WM PD values. This shows that PD maps were sensitive to the subtle changes in NAWM surrounding the lesions.

Discovery of KRB-456, a KRAS G12D Switch-I/II Allosteric Pocket Binder That Inhibits the Growth of Pancreatic Cancer Patient-derived Tumors.

Currently, there are no clinically approved drugs that directly thwart mutant KRAS G12D, a major driver of human cancer. Here, we report on the discovery of a small molecule, KRB-456, that binds KRAS G12D and inhibits the growth of pancreatic cancer patient-derived tumors. Protein nuclear magnetic resonance studies revealed that KRB-456 binds the GDP-bound and GCP-bound conformation of KRAS G12D by forming interactions with a dynamic allosteric binding pocket within the switch-I/II region. Isothermal titration calorimetry demonstrated that KRB-456 binds potently to KRAS G12D with 1.5-, 2-, and 6-fold higher affinity than to KRAS G12V, KRAS wild-type, and KRAS G12C, respectively. KRB-456 potently inhibits the binding of KRAS G12D to the RAS-binding domain (RBD) of RAF1 as demonstrated by GST-RBD pulldown and AlphaScreen assays. Treatment of KRAS G12D-harboring human pancreatic cancer cells with KRB-456 suppresses the cellular levels of KRAS bound to GTP and inhibits the binding of KRAS to RAF1. Importantly, KRB-456 inhibits P-MEK, P-AKT, and P-S6 levels in vivo and inhibits the growth of subcutaneous and orthotopic xenografts derived from patients with pancreatic cancer whose tumors harbor KRAS G12D and KRAS G12V and who relapsed after chemotherapy and radiotherapy. These results warrant further development of KRB-456 for pancreatic cancer. SIGNIFICANCE: There are no clinically approved drugs directly abrogating mutant KRAS G12D. Here, we discovered a small molecule, KRB-456, that binds a dynamic allosteric binding pocket within the switch-I/II region of KRAS G12D. KRB-456 inhibits P-MEK, P-AKT, and P-S6 levels in vivo and inhibits the growth of subcutaneous and orthotopic xenografts derived from patients with pancreatic cancer. This discovery warrants further advanced preclinical and clinical studies in pancreatic cancer.

Fetal-placental MR angiography at 1.5 T and 3 T.

OBJECTIVES: The objective of this work was to investigate the application of 2D Time-of-Flight (TOF) magnetic resonance angiography (MRA) to observe the placental vasculature at both 1.5 T and 3 T. METHODS: Fifteen appropriate for gestational age (AGA) (GA: 29.7 ± 3.4 weeks; GA range: 23 and 6/7 weeks to 36 and 2/7 weeks) and eleven patients with an abnormal singleton pregnancy (GA: 31.4 ± 4.4 weeks; GA range: 24 weeks to 35 and 2/7 weeks) were recruited in the study. Three AGA patients were scanned twice at different gestational ages. Patients were scanned either at 3 T or 1.5 T using both T2-HASTE and 2D TOF to image the entire placental vasculature. RESULTS: The umbilical, chorionic vessels, stem vessels, arcuate arteries, radial arteries, and spiral arteries were shown in most of the subjects. Hyrtl's anastomosis was found in two subjects in the 1.5 T data. The uterine arteries were observed in more than half of the subjects. For those patients scanned twice, the same spiral arteries were identified in both scans. CONCLUSIONS: 2D TOF is a technique that can be applied in studying the fetal-placental vasculature at both 1.5 T and 3 T.

Four-way classification of Alzheimer's disease using deep Siamese convolutional neural network with triplet-loss function.

Alzheimer's disease (AD) is a neurodegenerative disease that causes irreversible damage to several brain regions, including the hippocampus causing impairment in cognition, function, and behaviour. Early diagnosis of the disease will reduce the suffering of the patients and their family members. Towards this aim, in this paper, we propose a Siamese Convolutional Neural Network (SCNN) architecture that employs the triplet-loss function for the representation of input MRI images as k-dimensional embeddings. We used both pre-trained and non-pretrained CNNs to transform images into the embedding space. These embeddings are subsequently used for the 4-way classification of Alzheimer's disease. The model efficacy was tested using the ADNI and OASIS datasets which produced an accuracy of 91.83% and 93.85%, respectively. Furthermore, obtained results are compared with similar methods proposed in the literature.

Radioiodine Therapy in Patient with Differentiated Thyroid Cancer and End-Stage Renal Disease on Maintenance Hemodialysis: Case Report with Review of Literature.

Surgical resection followed by radioactive iodine (131I) therapy constitutes a standard treatment for differentiated thyroid cancer. 131I is normally excreted through the kidneys, and treatment of patients with end-stage renal disease on hemodialysis requires special attention to the dose of 131I, the timing of dialysis, and radiation safety. We present a case of end-stage renal disease in a postthyroidectomy patient on hemodialysis who required radioactive iodine ablation, and we review the literature.

Fluoro-2-Deoxyglucose-Positron Emission Tomography/Computed Tomography in the Diagnosis and Management of Adrenocortical Carcinoma: A 10-Year Experience from a Tertiary Care Institute.

Purpose: Adrenocortical carcinoma (ACC) is a rare primary malignancy of the adrenal gland. The present study was aimed to compare the performance of fluoro-2-deoxyglucose-positron emission tomography-computed tomography (FDG-PET-CT) compared to contrast-enhanced computed tomography (CECT) in diagnosis and management of ACC. Materials and Methods: A retrospective analysis of the PET-CT studies from January 2010 to October 2020 was performed. Patients with adrenal lesions suspicious of ACC and diagnosed cases of ACC who underwent PET-CT for staging, restaging, and surveillance were reanalyzed. The PET-CT parameters were compared with the clinical, biochemical, histopathological, and CECT parameters. Results: The study included 96 scans performed in 77 patients (36 males, aged 40.4 ± 17.9 years). Of these, 55 scans were performed to diagnose and stage suspected ACC (30 of them diagnosed as ACC), 31 for restaging, and 10 scans for surveillance of ACC. PET/CT revealed metastases from an extra-adrenal primary in 5/55 patients. FDG-PET-CT had a sensitivity and specificity of 100% and 70% to diagnose ACC. Standardized uptake value-peak more than 5.4 had a sensitivity of 90.9% and specificity of 91.7% for differentiating ACC from non-ACC lesions, while tumor-to-liver ratio peak (TLRpeak) of 3.3 was most specific. PET-CT changed the staging in 23.3% of the patients with an accuracy of 100%. PET-CT changed the management plan in 25.8% of the patients during restaging with a sensitivity and specificity of 95.6% and 100%, respectively. For surveillance, CECT was as sensitive as PET-CT; however, PET-CT was more specific (100% vs. 97.9%). Conclusion: FDG-PET-CT performs better than CECT in the diagnosis, staging, restaging, and surveillance of ACC.

Angiogenesis-Targeted 68Ga-DOTA-RGD2 PET/CT Imaging: a Potential Theranostic Application in the Case of Chondrosarcoma.

Chondrosarcoma is a cartilaginous tumor of mesenchymal origin. The histology and grade of the tumor determine the chances of relapse and survival. These tumors usually respond poorly to chemo-radiotherapy in cases of non-resectable and recurrent disease. 18F-FDG PET/CT has been used in evaluation of recurrence. However, these tumors show only mild to moderate FDG avidity due to their lower mitotic activity and large acellular matrix. These tumors are known to have a high degree of angiogenesis, especially in those of higher grade. We present a case of a 53-year-old man with grade II chondrosarcoma of the left femur showing only mild avidity on 18F-FDG PET/CT but showing moderate to intense tracer avidity on 68Ga-DOTA-RGD2 PET/CT. This may enable the use of angiogenesis-targeted positron and beta-emitting radiopharmaceuticals as a potentially new theranostic alternative treatment in cases of refractory metastatic chondrosarcoma.

Estimating cerebral venous oxygenation in human fetuses with ventriculomegaly using quantitative susceptibility mapping.

RATIONALE AND OBJECTIVES: The goal of this study was to estimate venous blood oxygen saturation (SvO2) in the superior sagittal sinus (SSS) in fetal brains with ventriculomegaly (VM) using quantitative susceptibility mapping (QSM). MATERIALS AND METHODS: A radiofrequency spoiled gradient echo sequence was used to evaluate data on 19 fetuses with VM (gestational age(GA): median = 29.9 weeks (range 23 to 37.3 weeks)) and 20 healthy fetuses (GA: median = 30.9 (range 22.7 to 38.7 weeks)) at 1.5 T. Susceptibility weighted images encompassing the entire fetal brain were acquired within 1 min. An iterative, geometry constraint-based thresholded k-space division algorithm was used for generating QSM data of the fetal brain. The venous oxygen saturation was calculated using the magnetic susceptibility of the SSS obtained from the QSM data. Mixed-model analysis of variance and interobserver variability assessment were used to analyze the results. RESULTS: The median SvO2 values in the entire VM cohort as well as for second and third trimester fetuses (with interquartile range) were: 67.8% (63.2%, 73.6%), 73.1% (69.1%, 77.3%) and 63.8% (59.4%, 68.1%), respectively. The corresponding median SvO2 value in the healthy control group was: 65.3% (58.3%, 68.2%), 67.5% (61.7%, 69.2%) and 60.8% (53.6%, 68.2%), respectively. However, the difference of SvO2 between VM and control groups was not significant at the p = 0.05 level (p = 0.076). The SvO2 was found decreasing significantly with GA in the healthy control group (p < 0.05). CONCLUSIONS: We report for the first time the estimation of cerebral SvO2 in human fetuses with VM using QSM. This measure of oxygen saturation might be beneficial in assessing and monitoring the metabolic status of the fetus in various clinical conditions.

Revealing vascular abnormalities and measuring small vessel density in multiple sclerosis lesions using USPIO.

BACKGROUND AND PURPOSE: Multiple Sclerosis (MS) is a progressive, inflammatory, neuro-degenerative disease of the central nervous system (CNS) characterized by a wide range of histopathological features including vascular abnormalities. In this study, an ultra-small superparamagnetic iron oxide (USPIO) contrast agent, Ferumoxytol, was administered to induce an increase in susceptibility for both arteries and veins to help better reveal the cerebral microvasculature. The purpose of this work was to examine the presence of vascular abnormalities and vascular density in MS lesions using high-resolution susceptibility weighted imaging (SWI). METHODS: Six subjects with relapsing remitting MS (RRMS, age = 47.3 ± 11.8 years with 3 females and 3 males) and fourteen age-matched healthy controls were scanned at 3 T with SWI acquired before and after the infusion of Ferumoxytol. Composite data was generated by registering the FLAIR data to the high resolution SWI data in order to highlight the vascular information in MS lesions. Both the central vein sign (CVS) and, a new measure, the multiple vessel sign (MVS) were identified, along with any vascular abnormalities, in the lesions on pre- and post-contrast SWI-FLAIR fusion data. The small vessel density within the periventricular normal-appearing white matter (NAWM) and the periventricular lesions were compared for all subjects. RESULTS: Averaged across two independent raters, a total of 530 lesions were identified across all patients. The total number of lesions with vascularity on pre- and post-contrast data were 287 and 488, respectively. The lesions with abnormal vascular behavior were broken up into following categories: small lesions appearing only at the vessel boundary; dilated vessels within the lesions; and developmental venous angiomas. These vessel abnormalities observed within lesions increased from 55 on pre-contrast data to 153 on post-contrast data. Finally, across all the patients, the periventricular lesional vessel density was significantly higher (p < 0.05) than that of the periventricular NAWM. CONCLUSIONS: By inducing a super-paramagnetic susceptibility in the blood using Ferumoxytol, the vascular abnormalities in the RRMS patients were revealed and small vessel densities were obtained. This approach has the potential to monitor the venous vasculature present in MS lesions, catalogue their characteristics and compare the vascular structures spatially to the presence of lesions. These enhanced vascular features may provide new insight into the pathophysiology of MS.

Ectopic Parathyroid Adenoma Mimicking as a Neuroendocrine Tumor on Ga68- DOTANOC Positron Emission Tomography/Computed Tomography Imaging.

Parathyroid adenoma sometimes present in ectopic location and may pose a difficulty in both diagnosis and localization. We report a case of a young lady suspected to have neuroendocrine tumor of the mediastinum demonstrating synaptophysin positivity on an initial core needle biopsy. Ga-68 DOTANOC positron emission tomography-computed tomography revealed a somatostatin receptor-expressing lesion in the anterior mediastinum with tracer avid multiple lytic bone lesions. On further biochemical and imaging workup with Tc-99 m SESTAMIBI, a diagnosis of ectopic parathyroid adenoma was made which was further confirmed with surgical excision.

Longitudinal Magnetic Resonance Imaging of Cerebral Microbleeds in Multiple Sclerosis Patients.

We hypothesized that cerebral microbleeds (CMBs) in multiple sclerosis (MS) patients will be detected with higher prevalence compared to healthy controls (HC) and that quantitative susceptibility mapping (QSM) will help remove false positives seen in susceptibility weighted imaging (SWI). A cohort of 100 relapsing remitting MS subjects scanned at 3T were used to validate a set of CMB detection guidelines specifically using QSM. A second longitudinal cohort of 112 MS and 25 HCs, also acquired at 3T, was reviewed across two time points. Both cohorts were imaged with SWI and fluid attenuated inversion recovery. Fourteen subjects in the first cohort (14%, 95% CI 8-21%) and twenty-one subjects in the second cohort (18.7%, 95% CI 11-27%) had at least one CMB. The combined information from SWI and QSM allowed us to discern stable CMBs and new CMBs from potential mimics and evaluate changes over time. The longitudinal results demonstrated that longer disease duration increased the chance to develop new CMBs. Higher age was also associated with increased CMB prevalence for MS and HC. We observed that MS subjects developed new CMBs between time points, indicating the need for longitudinal quantitative imaging of CMBs.

MicroRNAs in neuroblastoma tumorigenesis, therapy resistance, and disease evolution.

Neuroblastoma (NB) deriving from neural crest cells is the most common extra-cranial solid cancer at infancy. NB originates within the peripheral sympathetic ganglia in adrenal medulla and along the midline of the body. Clinically, NB exhibits significant heterogeneity stretching from spontaneous regression to rapid progression to therapy resistance. MicroRNAs (miRNAs, miRs) are small (19-22 nt in length) non-coding RNAs that regulate human gene expression at the post-transcriptional level and are known to regulate cellular signaling, growth, differentiation, death, stemness, and maintenance. Consequently, the function of miRs in tumorigenesis, progression and resistance is of utmost importance for the understanding of dysfunctional cellular pathways that lead to disease evolution, therapy resistance, and poor clinical outcomes. Over the last two decades, much attention has been devoted to understanding the functional roles of miRs in NB biology. This review focuses on highlighting the important implications of miRs within the context of NB disease progression, particularly miRs' influences on NB disease evolution and therapy resistance. In this review, we discuss the functions of both the "oncomiRs" and "tumor suppressor miRs" in NB progression/therapy resistance. These are the critical components to be considered during the development of novel miR-based therapeutic strategies to counter therapy resistance.

De novo regulation of RD3 synthesis in residual neuroblastoma cells after intensive multi-modal clinical therapy harmonizes disease evolution.

Most high-risk neuroblastomas that initially respond to therapy will ultimately relapse. Currently, no curative treatment is available. Acquired genetic/molecular rearrangement in therapy-resistant cells contributes to tumor relapse. Recently, we identified significant RD3 loss in progressive disease (PD) and defined its association with advanced disease-stage and poor clinical outcomes. Here, we investigated whether RD3 loss is an acquired process in cells that survive intensive multi-modal clinical therapy (IMCT) and its significance in disease evolution. RD3 status (mRNA, protein) during diagnosis (Dx) and PD after IMCT was investigated in NB patient cohort (n = 106), stage-4 NB cell lines (n = 15) with known treatment status and validated with independent data from another set of 15 cell-lines. Loss of RD3 in metastatic disease was examined using a mouse model of PD and metastatic-site-derived aggressive cells (MSDACs) ex vivo. RD3 silencing/expression assessed changes in metastatic state. Influence of RD3 loss in therapy resistance was examined through independent in vitro and in vivo studies. A significant loss of RD3 mRNA and protein was observed in resistant cells derived from patients with PD after IMCT. This is true to the effect within and between patients. Results from the mouse model identified significant transcriptional/translational loss of RD3 in metastatic tumors and MSDACs. RD3 re-expression in MSDACs and silencing RD3 in parental cells defined the functional relevance of RD3-loss in PD pathogenesis. Analysis of independent studies with salvage therapeutic agents affirmed RD3 loss in surviving resistant cells and residual tumors. The profound reductions in RD3 transcription indicate the de novo regulation of RD3 synthesis in resistant cells after IMCT. Defining RD3 loss in PD and the benefit of targeted reinforcement could improve salvage therapy for progressive neuroblastoma.

Retinal Degeneration Protein 3 (RD3) in normal human tissues: Novel insights.

The 195-amino-acid-long human Retinal Degeneration Protein 3 (RD3) is critical in the regulation of guanylate cyclase (GC) signaling and photoreceptor cell survival. Recently, we identified significant loss of RD3 in high-risk neuroblastoma and the influential role of RD3 in tumor progression. However, the functional characterization of RD3 in tumor systems has been hampered by the dearth of information on its localization in normal tissue and by the lack of antibodies suitable for staining FFPE tissue, primarily due to the inaccessibility of the epitopes. In this study, we validated a custom-synthesized RD3 antibody and investigated the expression/localization of RD3 in assorted human tissues. We observed stratified expression of RD3 in different cell types and subcellular location of retina. We demonstrated extensive positive RD3 immunoreactivity in various normal tissues and particularly strong dot-like perinuclear staining in the lining epithelial cells, suggesting that RD3 may play an important role in the normal functioning of epithelial cells. RD3 expression is limited in the CNS. While neuroblastoma is often RD3-positive, the adrenal medulla, where many neuroblastomas originate, is RD3-negative. Meta-analysis of RD3 transcriptional expression across normal tissues confirmed tissue-specific RD3 mRNA levels. Our results revealed the tissue-specific expression/localization profile of RD3 for the first time.

Insufficient (sub-native) helix content in soluble/solid aggregates of recombinant and engineered forms of IL-2 throws light on how aggregated IL-2 is biologically active.

Interleukin 2 (IL-2) is an extremely aggregation-prone, all-alpha helical cytokine. In its receptor-bound state, ~72 % of the polypeptide chain adopts helical structure and there is no beta sheet content whatsoever. In the past, recombinant IL-2 has been formulated and used therapeutically in humans, following production in E. coli. Therapeutic IL-2 consists entirely of functionally-active soluble aggregates with ~30 subunits per aggregate particle. Side-effects attributed to aggregation resulted in discontinuation of usage over a decade ago. Structurally, and biochemically, activity in IL-2 aggregates can potentially be explained in one of two ways : (a) individual IL-2 chains exist in sterically-accessible, receptor binding-competent (native) structures, allowing aggregates to bind directly to IL-2 receptors (IL-2R); alternatively, (b) IL-2 chains dissociate from aggregates, become free to adopt native structure, and then bind to IL-2R. We produced native IL-2 and numerous engineered forms in E. coli with the objective of obtaining insights into these possibilities. Each IL-2 variant was subjected to size exclusion chromatography, circular dichroism (CD) and Fourier transform infrared spectroscopy (FTIR). All forms produced and studied (including those with native IL-2 sequences) turned out to aggregate and also display less than ~50 % helix content as well as significant beta sheet content. No conditions were found that obviate aggregation. Aggregated IL-2 is thus insufficiently native-like to bind to IL-2R. Activity in aggregates thus probably owes to adoption of receptor binding-competent structures by chains that have already dissociated from aggregates.