Exploring the performance and explainability of fine-tuned BERT models for neuroradiology protocol assignment.

BACKGROUND: Deep learning has demonstrated significant advancements across various domains. However, its implementation in specialized areas, such as medical settings, remains approached with caution. In these high-stake environments, understanding the model's decision-making process is critical. This study assesses the performance of different pretrained Bidirectional Encoder Representations from Transformers (BERT) models and delves into understanding its decision-making within the context of medical image protocol assignment. METHODS: Four different pre-trained BERT models (BERT, BioBERT, ClinicalBERT, RoBERTa) were fine-tuned for the medical image protocol classification task. Word importance was measured by attributing the classification output to every word using a gradient-based method. Subsequently, a trained radiologist reviewed the resulting word importance scores to assess the model's decision-making process relative to human reasoning. RESULTS: The BERT model came close to human performance on our test set. The BERT model successfully identified relevant words indicative of the target protocol. Analysis of important words in misclassifications revealed potential systematic errors in the model. CONCLUSIONS: The BERT model shows promise in medical image protocol assignment by reaching near human level performance and identifying key words effectively. The detection of systematic errors paves the way for further refinements to enhance its safety and utility in clinical settings.

Chronic cough and an atypical pattern of peripheral pulmonary opacities: A case report secondary to suspected drug onset.

INTRODUCTION: Chronic eosinophilic pneumonia (CEP) is an idiopathic interstitial lung disease with nonspecific symptoms that involves a complex inflammatory cascade. CASE STUDY: A 36-year-old prisoner with a history of psoriasis presented with progressive worsening dyspnea, chest pain, and cough. His symptoms started 2-months after starting adalimumab, a tumor necrosis factor (TNF)-inhibitor, for psoriasis treatment. RESULTS: Initial workup revealed 27% eosinophils on complete blood count, elevated IgE levels on bronchoalveolar lavage, and bilateral peripheral lung opacities on imaging. The patient's symptoms and eosinophilia improved markedly after starting corticosteroids. Based on these findings, the patient was diagnosed with CEP. CONCLUSION: To our knowledge, this is the first case of asthma and CEP in a patient taking adalimumab. We suspect adalimumab unmasked the Th2 cell pathway response that was otherwise suppressed by psoriasis, a primarily Th1 cell pathway disease. The patient's pulmonary changes can be attributed to an eosinophilic asthma phenotype with adalimumab putatively indirectly causing CEP.

Pittsburgh Compound-B (PiB) binds amyloid ß-protein protofibrils.

The neuropathology of Alzheimer's disease (AD) includes amyloid plaque formation by the amyloid ß-protein (Aß) and intracellular paired helical filament formation by tau protein. These neuropathogenetic features correlate with disease progression and have been revealed in brains of AD patients using positron emission tomography (PET). One of the most useful positron emission tomography imaging agents has been Pittsburgh Compound-B (PiB). However, since its introduction in 2002, substantial evidence has accumulated suggesting that Aß oligomerization and protofibril formation, rather than fibril formation per se, may be the more important pathogenetic event in AD. Detecting protofibrils and oligomeric forms of Aß thus may be of value. We report here the results of experiments to determine whether PiB binds to oligomers or protofibrils formed by Aß40 and Aß42. We observed strong binding to Aß42 fibrils, significant binding to protofibrils, and weaker binding to Aß42 oligomers. PiB also binds Aß40 fibrils, but its binding to Aß40 protofibrils and oligomers is substantially lower than for that observed for Aß42.

Design, Characterization, and Use of a Novel Amyloid ß-Protein Control for Assembly, Neurotoxicity, and Gene Expression Studies.

A key pathogenic agent in Alzheimer's disease (AD) is the amyloid ß-protein (Aß), which self-assembles into a variety of neurotoxic structures. Establishing structure-activity relationships for these assemblies, which is critical for proper therapeutic target identification and design, requires aggregation and neurotoxicity experiments that are properly controlled with respect to the Aß peptide itself. "Reverse" Aß or non-Aß peptides suffer from the fact that their biophysical properties are too similar or dissimilar, respectively, to those of native Aß for them to be appropriate controls. For this reason, we used simple protein design principles to create scrambled Aß peptides predicted to behave distinctly from native Aß. We showed that our prediction was true by monitoring secondary structure dynamics with thioflavin T fluorescence and circular dichroism spectroscopy, determining oligomer size distributions, and assaying neurotoxic activity. We then demonstrated the utility of the scrambled Aß peptides by using them to control experiments examining the effects of Aß monomers, dimers, higher-order oligomers, and fibrils on gene expression in primary rat hippocampal neurons. Significant changes in gene expression were observed for all peptide assemblies, but fibrils induced the largest changes. Weighted gene co-expression network analysis revealed two predominant gene modules related to Aß treatment. Many genes within these modules were associated with inflammatory signaling pathways.

Design and Characterization of Chemically Stabilized Aß42 Oligomers.

A popular working hypothesis of Alzheimer's disease causation is amyloid ß-protein oligomers are the key neuropathogenetic agents. Rigorously elucidating the role of oligomers requires the production of stable oligomers of each size. We previously used zero-length photochemical cross-linking to allow stabilization, isolation, and determination of structure-activity relationships of pure populations of Aß40 dimers, trimers, and tetramers. We also attempted to study Aß42 but found that Aß42 oligomers subjected to the same procedures were not completely stable. On the basis of the fact that Tyr is a critical residue in cross-linking chemistry, we reasoned that the chemical accessibility of Tyr10 in Aß42 must differ from that in Aß40. We thus chemically synthesized four singly substituted Tyr variants that placed the Tyr in different positions across the Aß42 sequence. We then studied the stability of the resulting cross-linked oligomers as well as procedures for fractionating the oligomers to obtain pure populations of different sizes. We found that [Phe(10),Tyr(42)]Aß42 produced stable oligomers yielding highly pure populations of dimers through heptamers. This provides the means to establish formal structure-activity relationships of these important Aß42 assemblies. In addition, we were able to analyze the dissociation patterns of non-cross-linked oligomers to produce a model for oligomer formation. This work is relevant to the determination of structure-activity relationships that have the potential to provide mechanistic insights into disease pathogenesis.

Inhibiting amyloid ß-protein assembly: Size-activity relationships among grape seed-derived polyphenols.

Epidemiological evidence that red wine consumption negatively correlates with risk of Alzheimer's disease has led to experimental studies demonstrating that grape seed extracts inhibit the aggregation and oligomerization of Aß in vitro and ameliorate neuropathology and behavioral deficits in a mouse model of Alzheimer's disease. The active agent in the extracts is a mixed population of polyphenolic compounds. To evaluate the relative potency of each of these compounds, HPLC was used to fractionate the mixture into monomers, dimers, and oligomers. Each fraction was analyzed for its effect on Aß conformational dynamics (circular dichroism), oligomerization (zero-length photochemical cross-linking), aggregation kinetics (Thioflavin T fluorescence), and morphology (electron microscopy). The relative activities of each fraction were determined on the basis of molar concentration (mol/L) or mass concentration (g/L). When molar concentration, the number concentration of each polyphenolic compound, was considered, the oligomer fraction was the most potent inhibitor of Aß oligomerization and aggregation. However, when mass concentration, the number concentration of phenolic groups, was considered, monomers were the most potent inhibitors. To understand these ostensibly contradictory results, a model of polyphenol:Aß complexation was developed. This model, which was found to be consistent with published X-ray crystallographic studies, offers an explanation for the effects of functional group polyvalency on inhibitor activity. Our data emphasize the importance of an in-depth understanding of the mechanism(s) underlying 'concentration dependence' in inhibitor systems involving polyfunctional agents.

Arterial Spin-Labeling and DSC Perfusion Metrics Improve Agreement in Neuroradiologists' Clinical Interpretations of Posttreatment High-Grade Glioma Surveillance MR Imaging-An Institutional Experience.

BACKGROUND AND PURPOSE: MR perfusion has shown value in the evaluation of posttreatment high-grade gliomas, but few studies have shown its impact on the consistency and confidence of neuroradiologists' interpretation in routine clinical practice. We evaluated the impact of adding MR perfusion metrics to conventional contrast-enhanced MR imaging in posttreatment high-grade glioma surveillance imaging. MATERIALS AND METHODS: This retrospective study included 45 adults with high-grade gliomas who had posttreatment perfusion MR imaging. Four neuroradiologists assigned Brain Tumor Reporting and Data System scores for each examination on the basis of the interpretation of contrast-enhanced MR imaging and then after the addition of arterial spin-labeling-CBF, DSC-relative CBV, and DSC-fractional tumor burden. Interrater agreement and rater agreement with a multidisciplinary consensus group were assessed with κ statistics. Raters used a 5-point Likert scale to report confidence scores. The frequency of clinically meaningful score changes resulting from the addition of each perfusion metric was determined. RESULTS: Interrater agreement was moderate for contrast-enhanced MR imaging alone (κ = 0.63) and higher with perfusion metrics (arterial spin-labeling-CBF, κ = 0.67; DSC-relative CBV, κ = 0.66; DSC-fractional tumor burden, κ = 0.70). Agreement between raters and consensus was highest with DSC-fractional tumor burden (κ = 0.66-0.80). Confidence scores were highest with DSC-fractional tumor burden. Across all raters, the addition of perfusion resulted in clinically meaningful interpretation changes in 2%-20% of patients compared with contrast-enhanced MR imaging alone. CONCLUSIONS: Adding perfusion to contrast-enhanced MR imaging improved interrater agreement, rater agreement with consensus, and rater confidence in the interpretation of posttreatment high-grade glioma MR imaging, with the highest agreement and confidence scores seen with DSC-fractional tumor burden. Perfusion MR imaging also resulted in interpretation changes that could change therapeutic management in up to 20% of patients.

Fibrodysplasia ossificans progressiva (FOP) presenting as a rapidly growing non-calcified neck mass.

Fibrodysplasia ossificans progressiva (FOP) is an extremely rare autosomal dominant inherited disorder leading to mature ossification within soft tissues. We report a 62-year-old female with a 3-week history of a rapidly enlarging left neck mass with no associated symptoms. A neck CT showed a ~10 cm solid-appearing non-calcified left neck mass that markedly decreased in size on a one-month follow-up neck MRI, but with new extensive edema/intense enhancement in floor of the mouth. Prior radiographs documented hallux valgus and heterotopic ossification of the psoas/paraspinal muscles and shoulder girdle. In this case of FOP, no intervention was implemented and the symptoms improved over time and thus paralleled other such cases for flare-ups. Clinicians should be aware of this rare entity, as it is frequently misdiagnosed as cancer or other benign entities such as infection, resulting in biopsies that can often hasten disease progression.

Assessment of Motor Dysfunction with Virtual Reality in Patients Undergoing [123I]FP-CIT SPECT/CT Brain Imaging.

[123I]FP-CIT SPECT has been valuable for distinguishing Parkinson disease (PD) from essential tremor. However, its performance for quantitative assessment of motor dysfunction has not been established. A virtual reality (VR) application was developed and compared with [123I]FP-CIT SPECT/CT for detection of severity of motor dysfunction. Forty-four patients (21 males, 23 females, age 64.5 ± 12.4) with abnormal [123I]FP-CIT SPECT/CT underwent assessment of bradykinesia, activities of daily living, and tremor with VR. Support vector machines (SVM) machine learning models were applied to VR and SPECT data. Receiver operating characteristic (ROC) analysis demonstrated greater area under the curve (AUC) for VR (0.8418, 95% CI 0.6071-0.9617) compared with brain SPECT (0.5357, 95% CI 0.3373-0.7357, p = 0.029) for detection of motor dysfunction. Logistic regression identified VR as an independent predictor of motor dysfunction (Odds Ratio 326.4, SE 2.17, p = 0.008). SVM for prediction of the Unified Parkinson's Disease Rating Scale Part III (UPDRS-III) demonstrated greater R-squared of 0.713 (p = 0.008) for VR, compared with 0.0764 (p = 0.361) for brain SPECT. This study demonstrates that VR can be safely used in patients prior to [123I]FP-CIT SPECT imaging and may improve prediction of motor dysfunction. This test has the potential to provide a simple, objective, quantitative analysis of motor symptoms in PD patients.

Automated CT registration tool improves sensitivity to change in ventricular volume in patients with shunts and drains.

OBJECTIVE: CT is the mainstay imaging modality for assessing change in ventricular volume in patients with ventricular shunts or external ventricular drains (EVDs). We evaluated the performance of a novel fully automated CT registration and subtraction method to improve reader accuracy and confidence compared with standard CT. METHODS: In a retrospective evaluation of 49 ventricular shunt or EVD patients who underwent sequential head CT scans with an automated CT registration tool (CT CoPilot), three readers were assessed on their ability to discern change in ventricular volume between scans using standard axial CT images versus reformats and subtraction images generated by the registration tool. The inter-rater reliability among the readers was calculated using an intraclass correlation coefficient (ICC). Bland-Altman tests were performed to determine reader performance compared to semi-quantitative assessment using the bifrontal horn and third ventricular width. McNemar's test was used to determine whether the use of the registration tool increased the reader's level of confidence. RESULTS: Inter-rater reliability was higher when using the output of the registration tool (single measure ICC of 0.909 with versus 0.755 without the tool). Agreement between the readers' assessment of ventricular volume change and the semi-quantitative assessment improved with the registration tool (limits of agreement 4.1 vs 4.3). Furthermore, the tool improved reader confidence in determining increased or decreased ventricular volume (p < 0.001). CONCLUSION: Automated CT registration and subtraction improves the reader's ability to detect change in ventricular volume between sequential scans in patients with ventricular shunts or EVDs. ADVANCES IN KNOWLEDGE: Our automated CT registration and subtraction method may serve as a promising generalizable tool for accurate assessment of change in ventricular volume, which can significantly affect clinical management.

A peptide hairpin inhibitor of amyloid beta-protein oligomerization and fibrillogenesis.

Amyloid beta-protein (Abeta) self-assembly is linked strongly to Alzheimer's disease. We found that PP-Leu, a tridecapeptide analogue of broad-spectrum antiviral peptides termed theta-defensins, potently inhibits Abeta oligomer and fibril formation. This effect appeared to be mediated through sequestration of the amyloidogenic Abeta peptide in colloid-like assemblies. PP-Leu comprises a turn formed by a d-Pro-l-Pro amino acid dyad and stabilized by a disulfide bond, a motif that was exceptionally resistant to endoproteinase K digestion. This combination of assembly inhibitory activity and protease resistance suggests that PP-Leu may have potential therapeutic value.

NMDA receptor-dependent signaling pathways that underlie amyloid beta-protein disruption of LTP in the hippocampus.

Alzheimer's disease (AD), the most common neurodegenerative disease in the elderly population, is characterized by the hippocampal deposition of fibrils formed by amyloid beta-protein (A beta), a 40- to 42-amino-acid peptide. The folding of A beta into neurotoxic oligomeric, protofibrillar, and fibrillar assemblies is believed to mediate the key pathologic event in AD. The hippocampus is especially susceptible in AD and early degenerative symptoms include significant deficits in the performance of hippocampal-dependent cognitive abilities such as spatial learning and memory. Transgenic mouse models of AD that express C-terminal segments or mutant variants of amyloid precursor protein, the protein from which A beta is derived, exhibit age-dependent spatial memory impairment and attenuated long-term potentiation (LTP) in the hippocampal CA1 and dentate gyrus (DG) regions. Recent experimental evidence suggests that A beta disturbs N-methyl-D-aspartic acid (NMDA) receptor-dependent LTP induction in the CA1 and DG both in vivo and in vitro. Furthermore, these studies suggest that A beta specifically interferes with several major signaling pathways downstream of the NMDA receptor, including the Ca(2+)-dependent protein phosphatase calcineurin, Ca(2+)/calmodulin-dependent protein kinase II (CaMKII), protein phosphatase 1, and cAMP response element-binding protein (CREB). The influence of A beta on each of these downstream effectors of NMDA is reviewed in this article. Additionally, other mechanisms of LTP modulation, such as A beta attenuation of alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor currents, are briefly discussed.

Methionine oxidation, alpha-synuclein and Parkinson's disease.

The aggregation of normally soluble alpha-synuclein in the dopaminergic neurons of the substantia nigra is a crucial step in the pathogenesis of Parkinson's disease. Oxidative stress is believed to be a contributing factor in this disorder. Because it lacks Trp and Cys residues, mild oxidation of alpha-synuclein in vitro with hydrogen peroxide selectively converts all four methionine residues to the corresponding sulfoxides. Both oxidized and non-oxidized alpha-synucleins have similar unfolded conformations; however, the fibrillation of alpha-synuclein at physiological pH is completely inhibited by methionine oxidation. The inhibition results from stabilization of soluble oligomers of Met-oxidized alpha-synuclein. Furthermore, the Met-oxidized protein also inhibits fibrillation of unmodified alpha-synuclein. The degree of inhibition of fibrillation by Met-oxidized alpha-synuclein is proportional to the number of oxidized methionines. However, the presence of metals can completely overcome the inhibition of fibrillation of the Met-oxidized alpha-synuclein. Since oligomers of aggregated alpha-synuclein may be cytotoxic, these findings indicate that both oxidative stress and environmental metal pollution could play an important role in the aggregation of alpha-synuclein, and hence possibly Parkinson's disease. In addition, if the level of Met-oxidized alpha-synuclein was under the control of methionine sulfoxide reductase (Msr), then this could also be factor in the disease.

Characterization of oligomeric intermediates in alpha-synuclein fibrillation: FRET studies of Y125W/Y133F/Y136F alpha-synuclein.

The aggregation of alpha-synuclein is believed to be a critical step in the etiology of Parkinson's disease. A variety of biophysical techniques were used to investigate the aggregation and fibrillation of alpha-synuclein in which one of the four intrinsic Tyr residues was replaced by Trp, and two others by Phe, in order to permit fluorescence resonance energy transfer (FRET) between residues 39 (Tyr) and 125 (Trp). The mutant Y125W/Y133F/Y136F alpha-synuclein (one Tyr, one Trp) showed fibrillation kinetics similar to that of the wild-type, as did the Y125F/Y133F/Y136F (one Tyr, no Trp) and Y39F/Y125W/Y133F/Y136F (no Tyr, one Trp) mutants. Time-dependent changes in FRET, Fourier transform infrared, Trp fluorescence, dynamic light-scattering and other probes, indicate the existence of a transient oligomer, whose population reaches a maximum at the end of the lag time. This oligomer, in which the alpha-synuclein is in a partially folded conformation, is subsequently converted into fibrils, and has physical properties that are distinct from those of the monomer and fibrils. In addition, another population of soluble oligomers was observed to coexist with fibrils at completion of the reaction. The average distance between Tyr39 and Trp125 decreases from 24.9A in the monomer to 21.9A in the early oligomer and 18.8A in the late oligomer. Trp125 remains solvent-exposed in both the oligomers and fibrils, indicating that the C-terminal domain is not part of the fibril core. No FRET was observed in the fibrils, due to quenching of Tyr39 fluorescence in the fibril core. Thus, aggregation of alpha-synuclein involves multiple oligomeric intermediates and competing pathways.

Demonstration of Non-Gaussian Restricted Diffusion in Tumor Cells Using Diffusion Time-Dependent Diffusion-Weighted Magnetic Resonance Imaging Contrast.

The diffusion-weighted magnetic resonance imaging (DWI) technique enables quantification of water mobility for probing microstructural properties of biological tissue and has become an effective tool for collecting information about the underlying pathology of cancerous tissue. Measurements using multiple b-values have indicated biexponential signal attenuation, ascribed to "fast" (high ADC) and "slow" (low ADC) diffusion components. In this empirical study, we investigate the properties of the diffusion time (Δ)-dependent components of the diffusion-weighted (DW) signal in a constant b-value experiment. A xenograft gliobastoma mouse was imaged using Δ = 11 ms, 20 ms, 40 ms, 60 ms, and b = 500-4000 s/mm(2) in intervals of 500 s/mm(2). Data were corrected for EPI distortions, and the Δ-dependence on the DW-signal was measured within three regions of interest [intermediate- and high-density tumor regions and normal-appearing brain (NAB) tissue regions]. In this study, we verify the assumption that the slow decaying component of the DW-signal is non-Gaussian and dependent on Δ, consistent with restricted diffusion of the intracellular space. As the DW-signal is a function of Δ and is specific to restricted diffusion, manipulating Δ at constant b-value (cb) provides a complementary and direct approach for separating the restricted from the hindered diffusion component. We found that Δ-dependence is specific to the tumor tissue signal. Based on an extended biexponential model, we verified the interpretation of the diffusion time-dependent contrast and successfully estimated the intracellular restricted ADC, signal volume fraction, and cell size within each ROI.

Voxel Level Radiologic-Pathologic Validation of Restriction Spectrum Imaging Cellularity Index with Gleason Grade in Prostate Cancer.

PURPOSE: Restriction spectrum imaging (RSI-MRI), an advanced diffusion imaging technique, can potentially circumvent current limitations in tumor conspicuity, in vivo characterization, and location demonstrated by multiparametric magnetic resonance imaging (MP-MRI) techniques in prostate cancer detection. Prior reports show that the quantitative signal derived from RSI-MRI, the cellularity index, is associated with aggressive prostate cancer as measured by Gleason grade (GG). We evaluated the reliability of RSI-MRI to predict variance with GG at the voxel-level within clinically demarcated prostate cancer regions. EXPERIMENTAL DESIGN: Ten cases were processed using whole mount sectioning after radical prostatectomy. Regions of tumor were identified by an uropathologist. Stained prostate sections were scanned at high resolution (75 µm/pixel). A grid of tiles corresponding to voxel dimensions was graded using the GG system. RSI-MRI cellularity index was calculated from presurgical prostate MR scans and presented as normalized z-score maps. In total, 2,795 tiles were analyzed and compared with RSI-MRI cellularity. RESULTS: RSI-MRI cellularity index was found to distinguish between prostate cancer and benign tumor (t = 25.48, P < 0.00001). Significant differences were also found between benign tissue and prostate cancer classified as low-grade (GG = 3; t = 11.56, P < 0.001) or high-grade (GG ≥ 4; t = 24.03, P < 0.001). Furthermore, RSI-MRI differentiated between low and high-grade prostate cancer (t = 3.23; P = 0.003). CONCLUSIONS: Building on our previous findings of correlation between GG and the RSI-MRI among whole tumors, our current study reveals a similar correlation at voxel resolution within tumors. Because it can detect variations in tumor grade with voxel-level precision, RSI-MRI may become an option for planning targeted procedures where identifying the area with the most aggressive disease is important. Clin Cancer Res; 22(11); 2668-74. ©2016 AACR.

Effects of nitration on the structure and aggregation of alpha-synuclein.

Substantial evidence suggests that the aggregation of the presynaptic protein alpha-synuclein is a key step in the etiology of Parkinson's disease (PD). Although the molecular mechanisms underlying alpha-synuclein aggregation remain unknown, oxidative stress has been implicated in the pathogenesis of PD. Here, we report the effects of tyrosine nitration on the propensity of human recombinant alpha-synuclein to fibrillate in vitro. The properties of nitrated alpha-synuclein were investigated using a variety of biophysical and biochemical techniques, which revealed that nitration led to formation of a partially folded conformation with increased secondary structure relative to the intrinsically disordered structure of the monomer, and to oligomerization at neutral pH. The degree of self-association was concentration-dependent, but at 1 mg/mL, nitrated alpha-synuclein was predominantly an octamer. At low pH, small-angle X-ray scattering data indicated that the nitrated protein was monomeric. alpha-Synuclein fibrillation at neutral pH was completely inhibited by nitrotyrosination and is attributed to the formation of stable soluble oligomers. The presence of heparin or metals did not overcome the inhibition; however, the inhibitory effect was eliminated at low pH. The addition of nitrated alpha-synuclein inhibited fibrillation of non-modified alpha-synuclein at neutral pH. Potential implications of these findings to the etiology of Parkinson's disease are discussed.

Nitration inhibits fibrillation of human alpha-synuclein in vitro by formation of soluble oligomers.

The aggregation of alpha-synuclein in dopaminergic neurons is a critical factor in the etiology of Parkinson's disease (PD). Oxidative and nitrative stress is also implicated in PD. We examined the effect of nitration on the propensity of alpha-synuclein to fibrillate in vitro. Fibril formation of alpha-synuclein was completely inhibited by nitration, due to the formation of stable soluble oligomers (apparently octamers). More importantly the presence of sub-stoichiometric concentrations of nitrated alpha-synuclein led to inhibition of fibrillation of non-modified alpha-synuclein. These observations suggest that nitration of soluble alpha-synuclein may be a protective factor in PD, rather than a causative one.