Using neural networks to autonomously assess adequacy in intraoperative cholangiograms.

BACKGROUND: Intraoperative cholangiography (IOC) is a contrast-enhanced X-ray acquired during laparoscopic cholecystectomy. IOC images the biliary tree whereby filling defects, anatomical anomalies and duct injuries can be identified. In Australia, IOC are performed in over 81% of cholecystectomies compared with 20 to 30% internationally (Welfare AIoHa in Australian Atlas of Healthcare Variation, 2017). In this study, we aim to train artificial intelligence (AI) algorithms to interpret anatomy and recognise abnormalities in IOC images. This has potential utility in (a) intraoperative safety mechanisms to limit the risk of missed ductal injury or stone, (b) surgical training and coaching, and (c) auditing of cholangiogram quality. METHODOLOGY: Semantic segmentation masks were applied to a dataset of 1000 cholangiograms with 10 classes. Classes corresponded to anatomy, filling defects and the cholangiogram catheter instrument. Segmentation masks were applied by a surgical trainee and reviewed by a radiologist. Two convolutional neural networks (CNNs), DeeplabV3+ and U-Net, were trained and validated using 900 (90%) labelled frames. Testing was conducted on 100 (10%) hold-out frames. CNN generated segmentation class masks were compared with ground truth segmentation masks to evaluate performance according to a pixel-wise comparison. RESULTS: The trained CNNs recognised all classes.. U-Net and DeeplabV3+ achieved a mean F1 of 0.64 and 0.70 respectively in class segmentation, excluding the background class. The presence of individual classes was correctly recognised in over 80% of cases. Given the limited local dataset, these results provide proof of concept in the development of an accurate and clinically useful tool to aid in the interpretation and quality control of intraoperative cholangiograms. CONCLUSION: Our results demonstrate that a CNN can be trained to identify anatomical structures in IOC images. Future performance can be improved with the use of larger, more diverse training datasets. Implementation of this technology may provide cholangiogram quality control and improve intraoperative detection of ductal injuries or ductal injuries.

Effects of a comprehensive brain computed tomography deep learning model on radiologist detection accuracy.

OBJECTIVES: Non-contrast computed tomography of the brain (NCCTB) is commonly used to detect intracranial pathology but is subject to interpretation errors. Machine learning can augment clinical decision-making and improve NCCTB scan interpretation. This retrospective detection accuracy study assessed the performance of radiologists assisted by a deep learning model and compared the standalone performance of the model with that of unassisted radiologists. METHODS: A deep learning model was trained on 212,484 NCCTB scans drawn from a private radiology group in Australia. Scans from inpatient, outpatient, and emergency settings were included. Scan inclusion criteria were age ≥ 18 years and series slice thickness ≤ 1.5 mm. Thirty-two radiologists reviewed 2848 scans with and without the assistance of the deep learning system and rated their confidence in the presence of each finding using a 7-point scale. Differences in AUC and Matthews correlation coefficient (MCC) were calculated using a ground-truth gold standard. RESULTS: The model demonstrated an average area under the receiver operating characteristic curve (AUC) of 0.93 across 144 NCCTB findings and significantly improved radiologist interpretation performance. Assisted and unassisted radiologists demonstrated an average AUC of 0.79 and 0.73 across 22 grouped parent findings and 0.72 and 0.68 across 189 child findings, respectively. When assisted by the model, radiologist AUC was significantly improved for 91 findings (158 findings were non-inferior), and reading time was significantly reduced. CONCLUSIONS: The assistance of a comprehensive deep learning model significantly improved radiologist detection accuracy across a wide range of clinical findings and demonstrated the potential to improve NCCTB interpretation. CLINICAL RELEVANCE STATEMENT: This study evaluated a comprehensive CT brain deep learning model, which performed strongly, improved the performance of radiologists, and reduced interpretation time. The model may reduce errors, improve efficiency, facilitate triage, and better enable the delivery of timely patient care. KEY POINTS: • This study demonstrated that the use of a comprehensive deep learning system assisted radiologists in the detection of a wide range of abnormalities on non-contrast brain computed tomography scans. • The deep learning model demonstrated an average area under the receiver operating characteristic curve of 0.93 across 144 findings and significantly improved radiologist interpretation performance. • The assistance of the comprehensive deep learning model significantly reduced the time required for radiologists to interpret computed tomography scans of the brain.

ADMANI: Annotated Digital Mammograms and Associated Non-Image Datasets.

Supplemental material is available for this article. Keywords: Mammography, Screening, Convolutional Neural Network (CNN) Published under a CC BY 4.0 license. See also the commentary by Cadrin-Chênevert in this issue.

Machine Learning Augmented Interpretation of Chest X-rays: A Systematic Review.

Limitations of the chest X-ray (CXR) have resulted in attempts to create machine learning systems to assist clinicians and improve interpretation accuracy. An understanding of the capabilities and limitations of modern machine learning systems is necessary for clinicians as these tools begin to permeate practice. This systematic review aimed to provide an overview of machine learning applications designed to facilitate CXR interpretation. A systematic search strategy was executed to identify research into machine learning algorithms capable of detecting >2 radiographic findings on CXRs published between January 2020 and September 2022. Model details and study characteristics, including risk of bias and quality, were summarized. Initially, 2248 articles were retrieved, with 46 included in the final review. Published models demonstrated strong standalone performance and were typically as accurate, or more accurate, than radiologists or non-radiologist clinicians. Multiple studies demonstrated an improvement in the clinical finding classification performance of clinicians when models acted as a diagnostic assistance device. Device performance was compared with that of clinicians in 30% of studies, while effects on clinical perception and diagnosis were evaluated in 19%. Only one study was prospectively run. On average, 128,662 images were used to train and validate models. Most classified less than eight clinical findings, while the three most comprehensive models classified 54, 72, and 124 findings. This review suggests that machine learning devices designed to facilitate CXR interpretation perform strongly, improve the detection performance of clinicians, and improve the efficiency of radiology workflow. Several limitations were identified, and clinician involvement and expertise will be key to driving the safe implementation of quality CXR machine learning systems.

Charting the potential of brain computed tomography deep learning systems.

Brain computed tomography (CTB) scans are widely used to evaluate intracranial pathology. The implementation and adoption of CTB has led to clinical improvements. However, interpretation errors occur and may have substantial morbidity and mortality implications for patients. Deep learning has shown promise for facilitating improved diagnostic accuracy and triage. This research charts the potential of deep learning applied to the analysis of CTB scans. It draws on the experience of practicing clinicians and technologists involved in development and implementation of deep learning-based clinical decision support systems. We consider the past, present and future of the CTB, along with limitations of existing systems as well as untapped beneficial use cases. Implementing deep learning CTB interpretation systems and effectively navigating development and implementation risks can deliver many benefits to clinicians and patients, ultimately improving efficiency and safety in healthcare.

Do comprehensive deep learning algorithms suffer from hidden stratification? A retrospective study on pneumothorax detection in chest radiography.

OBJECTIVES: To evaluate the ability of a commercially available comprehensive chest radiography deep convolutional neural network (DCNN) to detect simple and tension pneumothorax, as stratified by the following subgroups: the presence of an intercostal drain; rib, clavicular, scapular or humeral fractures or rib resections; subcutaneous emphysema and erect versus non-erect positioning. The hypothesis was that performance would not differ significantly in each of these subgroups when compared with the overall test dataset. DESIGN: A retrospective case-control study was undertaken. SETTING: Community radiology clinics and hospitals in Australia and the USA. PARTICIPANTS: A test dataset of 2557 chest radiography studies was ground-truthed by three subspecialty thoracic radiologists for the presence of simple or tension pneumothorax as well as each subgroup other than positioning. Radiograph positioning was derived from radiographer annotations on the images. OUTCOME MEASURES: DCNN performance for detecting simple and tension pneumothorax was evaluated over the entire test set, as well as within each subgroup, using the area under the receiver operating characteristic curve (AUC). A difference in AUC of more than 0.05 was considered clinically significant. RESULTS: When compared with the overall test set, performance of the DCNN for detecting simple and tension pneumothorax was statistically non-inferior in all subgroups. The DCNN had an AUC of 0.981 (0.976-0.986) for detecting simple pneumothorax and 0.997 (0.995-0.999) for detecting tension pneumothorax. CONCLUSIONS: Hidden stratification has significant implications for potential failures of deep learning when applied in clinical practice. This study demonstrated that a comprehensively trained DCNN can be resilient to hidden stratification in several clinically meaningful subgroups in detecting pneumothorax.

Distal Medium Vessel Occlusions Can Be Accurately and Rapidly Detected Using Tmax Maps.

Background and Purpose: Distal medium vessel occlusions (DMVOs) are increasingly considered for endovascular thrombectomy but are difficult to detect on computed tomography angiography (CTA). We aimed to determine whether time-to-maximum of tissue residue function (Tmax) maps, derived from CT perfusion, can be used as a triage screening tool to accurately and rapidly identify patients with DMVOs. Methods: Consecutive code stroke patients who underwent multimodal CT were screened retrospectively. Two experienced readers evaluated all patients' Tmax maps in consensus for presence of delay in an arterial territory (territorial Tmax delay). The diagnostic accuracy of this surrogate for identifying DMVOs was determined using receiver-operating characteristic analysis. CTA, interpreted by 2 experienced neuroradiologists with access to all imaging data, served as the reference standard. Diagnostic performance of 4 other readers with different levels of experience for identifying DMVOs on Tmax versus CTA was also assessed. These readers independently assessed patients' Tmax maps and CTAs in 2 separate timed sessions, and areas under the receiver-operating characteristic curves were compared using the DeLong algorithm. The Wilcoxon signed-rank test was used to comparatively assess diagnostic speed. Results: Three hundred seventy-three code stroke patients (median age, 70 years; 56% male, 70 with a DMVO) were included. Territorial Tmax delay had a sensitivity of 100% (CI95, 94.9%­100%) and specificity of 87.8% (CI95, 83.6%­91.3%) for presence of a DMVO, yielding an area under the receiver-operating characteristic curves of 0.939 (CI95, 0.920­0.957). All 4 readers achieved sensitivity >95% and specificity >84% for detecting DMVOs using Tmax maps, with diagnostic accuracy (area under the receiver-operating characteristic curves) and speed that were significantly (P<0.001) higher than on CTA. Conclusions: Territorial Tmax delay had perfect sensitivity and high specificity for a DMVO. Tmax maps were accurately and rapidly interpreted by even inexperienced readers, and causes of false positives are easy to recognize and dismiss. These findings encourage the use of Tmax to identify patients with DMVOs.

Effect of a comprehensive deep-learning model on the accuracy of chest x-ray interpretation by radiologists: a retrospective, multireader multicase study.

BACKGROUND: Chest x-rays are widely used in clinical practice; however, interpretation can be hindered by human error and a lack of experienced thoracic radiologists. Deep learning has the potential to improve the accuracy of chest x-ray interpretation. We therefore aimed to assess the accuracy of radiologists with and without the assistance of a deep-learning model. METHODS: In this retrospective study, a deep-learning model was trained on 821 681 images (284 649 patients) from five data sets from Australia, Europe, and the USA. 2568 enriched chest x-ray cases from adult patients (≥16 years) who had at least one frontal chest x-ray were included in the test dataset; cases were representative of inpatient, outpatient, and emergency settings. 20 radiologists reviewed cases with and without the assistance of the deep-learning model with a 3-month washout period. We assessed the change in accuracy of chest x-ray interpretation across 127 clinical findings when the deep-learning model was used as a decision support by calculating area under the receiver operating characteristic curve (AUC) for each radiologist with and without the deep-learning model. We also compared AUCs for the model alone with those of unassisted radiologists. If the lower bound of the adjusted 95% CI of the difference in AUC between the model and the unassisted radiologists was more than -0·05, the model was considered to be non-inferior for that finding. If the lower bound exceeded 0, the model was considered to be superior. FINDINGS: Unassisted radiologists had a macroaveraged AUC of 0·713 (95% CI 0·645-0·785) across the 127 clinical findings, compared with 0·808 (0·763-0·839) when assisted by the model. The deep-learning model statistically significantly improved the classification accuracy of radiologists for 102 (80%) of 127 clinical findings, was statistically non-inferior for 19 (15%) findings, and no findings showed a decrease in accuracy when radiologists used the deep-learning model. Unassisted radiologists had a macroaveraged mean AUC of 0·713 (0·645-0·785) across all findings, compared with 0·957 (0·954-0·959) for the model alone. Model classification alone was significantly more accurate than unassisted radiologists for 117 (94%) of 124 clinical findings predicted by the model and was non-inferior to unassisted radiologists for all other clinical findings. INTERPRETATION: This study shows the potential of a comprehensive deep-learning model to improve chest x-ray interpretation across a large breadth of clinical practice. FUNDING: Annalise.ai.

Evaluation of deep learning-based artificial intelligence techniques for breast cancer detection on mammograms: Results from a retrospective study using a BreastScreen Victoria dataset.

INTRODUCTION: This study aims to evaluate deep learning (DL)-based artificial intelligence (AI) techniques for detecting the presence of breast cancer on a digital mammogram image. METHODS: We evaluated several DL-based AI techniques that employ different approaches and backbone DL models and tested the effect on performance of using different data-processing strategies on a set of digital mammographic images with annotations of pathologically proven breast cancer. RESULTS: Our evaluation uses the area under curve (AUC) and accuracy (ACC) for performance measurement. The best evaluation result, based on 349 test cases (930 test images), was an AUC of 0.8979 [95% confidence interval (CI) 0.873, 0.923] and ACC of 0.8178 [95% CI 0.785, 0.850]. This was achieved by an AI technique that utilises a certain family of DL models, namely ResNet, as its backbone, combines the global features extracted from the whole mammogram and the local features extracted from the automatically detected cancer and non-cancer local regions in the whole image, and leverages background cropping and text removal, contrast adjustment and more training data. CONCLUSION: DL-based AI techniques have shown promising results in retrospective studies for many medical image analysis applications. Our study demonstrates a significant opportunity to boost the performance of such techniques applied to breast cancer detection by exploring different types of approaches, backbone DL models and data-processing strategies. The promising results we have obtained suggest further development of AI reading services could transform breast cancer screening in the future.

Incidental detection of prostate cancer with computed tomography scans.

Prostate cancer (PCa) is the second most frequent type of cancer found in men worldwide, with around one in nine men being diagnosed with PCa within their lifetime. PCa often shows no symptoms in its early stages and its diagnosis techniques are either invasive, resource intensive, or has low efficacy, making widespread early detection onerous. Inspired by the recent success of deep convolutional neural networks (CNN) in computer aided detection (CADe), we propose a new CNN based framework for incidental detection of clinically significant prostate cancer (csPCa) in patients who had a CT scan of the abdomen/pelvis for other reasons. While CT is generally considered insufficient to diagnose PCa due to its inferior soft tissue characterisation, our evaluations on a relatively large dataset consisting of 139 clinically significant PCa patients and 432 controls show that the proposed deep neural network pipeline can detect csPCa patients at a level that is suitable for incidental detection. The proposed pipeline achieved an area under the receiver operating characteristic curve (ROC-AUC) of 0.88 (95% Confidence Interval: 0.86-0.90) at patient level csPCa detection on CT, significantly higher than the AUCs achieved by two radiologists (0.61 and 0.70) on the same task.

Sequences show rapid motor transfer and spatial translation in the oculomotor system.

Every day we perform learnt sequences of actions that seem to happen almost without awareness. It has been argued that for learning such sequences parallel learning networks exist - one using spatial coordinates and one using motor coordinates - with sequence acquisition involving a progressive shift from the former to the latter as a sequence is rehearsed. When sequences are interrupted by an out-of-sequence target, there is a delay in the response to the target, and so here we transiently interrupt oculomotor sequences to probe the influence of oculomotor rehearsal and spatial coordinates in sequence acquisition. For our main experiments, we used a repeating sequences of eight targets in length that was first learnt either using saccadic eye movements (left/right), manual responses (left/right or up/down) or as a sequence of colour (blue/red) requiring no motor response. The sequence was immediately repeated for saccadic eye movements, during which the influence of on out-of-sequence target (an interruption) was assessed. When a sequence is learnt beforehand in an abstract way (for example, as a sequence of colours or of orthogonally mapped manual responses), interruptions are immediately disruptive to latency, suggesting neither motor rehearsal nor specific spatial coordinates are essential for encoding sequences of actions and that sequences - no matter how they are encoded - can be rapidly translated into oculomotor coordinates. The magnitude of a disruption does, however, correspond to how well a sequence is learnt: introducing an interruption to an extended sequence before it was reliably learnt reduces the magnitude of the latency disruption.

Target direction rather than position determines oculomotor expectation in repeating sequences.

Saccadic latencies to targets appearing to the left and right of fixation in a repeating sequence are significantly increased when a target is presented out of sequence. Is this because the target is in the wrong position, the wrong direction, or both? To find out, we arranged for targets in a horizontal plane occasionally to appear with an unexpected eccentricity, though in the correct direction. This had no significant effect on latency, unlike what is observed when targets appeared in the unexpected direction. That subjects learnt sequences of directions rather than simply positions was further confirmed in an experiment where saccade direction was a repeating sequence, but eccentricity was randomised. Latency was elevated when a target was episodically presented in an unexpected direction. Latencies were also elevated when targets appeared in the correct hemifield but at an unexpected direction (35° polar angular displacement from the horizontal, a displacement roughly equivalent in collicular spacing to our unexpected eccentricity), although this elevation was of a smaller magnitude than when targets appeared in an unexpected direction along the horizontal. Finally, we confirmed that not all changes in the stimulus cause disruption: an unexpected change in the orientation or colour of the target did not alter latency. Our results show that in a repeating sequence, the oculomotor system is primarily concerned with predicting the direction of an upcoming eye movement rather than its position. This is consistent with models of oculomotor control developed for randomly appearing targets in which the direction and amplitude of saccades are programmed separately.

White matter changes in patients with Parkinson's disease carrying small CGG expansion FMR1 alleles: a pilot study.

BACKGROUND/AIMS: Alleles of the FMR1 gene containing small expansions of the CGG-trinucleotide repeat comprise premutation and grey-zone alleles. Premutation alleles may cause late-onset Fragile X-associated tremor/ataxia syndrome attributed to the neurotoxic effect of elevated FMR1 transcripts. Our earlier data suggested that both grey-zone and low-end premutation alleles might also play a significant role in the acquisition of the parkinsonian phenotype due to mitochondrial dysfunction caused by elevated FMR1 mRNA toxicity. These data were obtained through clinical and molecular comparisons between carriers of grey-zone/low-end premutation alleles and group-matched non-carrier controls from patients with idiopathic Parkinson's disease (iPD). We aimed to explore the relationship between grey-zone alleles, parkinsonism and white matter changes. METHODS: This study compared the extent and severity of white matter hyperintensity (WMH) on magnetic resonance imaging, using a semi-quantitative method, between 11 grey-zone/low-end premutation carriers and 20 non-carrier controls with iPD from our earlier study. Relationships between WMH scores, and cognitive and motor test scores were assessed for carriers and non-carriers. RESULTS: Supratentorial WMH scores, and tremor and ataxia motor scores were significantly higher in carriers compared with disease controls. Moreover, some associations between cognitive decline and WMH scores were specific for each respective carrier status category. CONCLUSIONS: The results support our earlier claim that grey-zone alleles contribute to the severity of parkinsonism and white matter changes.

Comparison of MRI sequences for evaluation of multiple sclerosis of the cervical spinal cord at 3 T.

PURPOSE: Debate remains regarding the utility of the traditional STIR (short inversion time recovery) sequence in aiding MRI diagnosis of spinal cord lesions in patients with multiple sclerosis (MS) and this sequence is not included in the current imaging guidelines. A recent study proposed a T1 weighted STIR as a superior alternative to the traditional STIR and T2 fast spin echo (FSE). Thus, the aim of this study was to compare the sensitivity of T2, standard STIR and T1 weighted STIR sequences in the evaluation of MS plaques on our 3 T system. METHODS AND MATERIALS: A retrospective analysis of patients with multiple sclerosis who presented to our institution over a period of 5 months and who had cervical cord lesions was undertaken. Patients had been examined with our institutional protocol which included T2 FSE, STIR and the recommended T1 STIR. Quantitative analysis of the lesions versus background cord using sample T-tests was performed for each sequence, and comparative analysis of the lesion contrast:background cord ratios of the 3 sequences (using two-way ANOVA tests) was performed. RESULTS: The T2 sequence was not as sensitive in detecting lesions versus the traditional STIR and T1 weighted STIR, with 10% of lesions not detected using statistical analysis (p<0.05). The traditional STIR also demonstrated greater contrast ratios than the T2 sequence (p<0.05) suggesting increased sensitivity. However, the T1 STIR demonstrated even greater contrast ratios than both the traditional STIR and T2 sequences (p<0.05). CONCLUSION: This study confirms earlier findings of the traditional STIRs increased sensitivity versus the T2 sequence. However, the new "T1 weighted STIR" appears to be even more sensitive than both these sequences showing potential promise as an alternative method to monitor demyelinating plaques of MS.

A state space based approach in non-linear hemodynamic response modeling with fMRI data.

In this paper we use the modified and integrated version of the balloon model in the analysis of fMRI data. We propose a new state space model realization for this balloon model and represent it with the standard A,B,C and D matrices widely used in system theory. A second order Padé approximation with equal numerator and denominator degree is used for the time delay approximation in the modeling of the cerebral blood flow. The results obtained through numerical solutions showed that the new state space model realization is in close agreement to the actual modified and integrated version of the balloon model. This new system theoretic formulation is likely to open doors to a novel way of analyzing fMRI data with real time robust estimators. With further development and validation, the new model has the potential to devise a generalized measure to make a significant contribution to improve the diagnosis and treatment of clinical scenarios where the brain functioning get altered. Concepts from system theory can readily be used in the analysis of fMRI data and the subsequent synthesis of filters and estimators.

Robust ODF smoothing for accurate estimation of fiber orientation.

Q-ball imaging was presented as a model free, linear and multimodal diffusion sensitive approach to reconstruct diffusion orientation distribution function (ODF) using diffusion weighted MRI data. The ODFs are widely used to estimate the fiber orientations. However, the smoothness constraint was proposed to achieve a balance between the angular resolution and noise stability for ODF constructs. Different regularization methods were proposed for this purpose. However, these methods are not robust and quite sensitive to the global regularization parameter. Although, numerical methods such as L-curve test are used to define a globally appropriate regularization parameter, it cannot serve as a universal value suitable for all regions of interest. This may result in over smoothing and potentially end up in neglecting an existing fiber population. In this paper, we propose to include an interpolation step prior to the spherical harmonic decomposition. This interpolation based approach is based on Delaunay triangulation provides a reliable, robust and accurate smoothing approach. This method is easy to implement and does not require other numerical methods to define the required parameters. Also, the fiber orientations estimated using this approach are more accurate compared to other common approaches.

Maximum likelihood estimation for Rician distributed data in analytical q-ball imaging.

Analytical q-ball imaging is widely used for reconstruction of orientation distribution function (ODF) using diffusion weighted MRI data. Estimating the spherical harmonic coefficients is a critical step in this method. Least squares (LS) is widely used for this purpose assuming the noise to be additive Gaussian. However, Rician noise is considered as a more appropriate model to describe noise in MR signal. Therefore, the current estimation techniques are valid only for high SNRs with Gaussian distribution approximating the Rician distribution. The aim of this study is to present an estimation approach considering the actual distribution of the data to provide reliable results particularly for the case of low SNR values. Maximum likelihood (ML) is investigated as a more effective estimation method. However, no closed form estimator is presented as the estimator becomes nonlinear for the noise assumption of the Rician distribution. Consequently, the results of LS estimator is used as an initial guess and the more refined answer is achieved using iterative numerical methods. According to the results, the ODFs reconstructed from low SNR data are in close agreement with ODFs reconstructed from high SNRs when Rician distribution is considered. Also, the error between the estimated and actual fiber orientations was compared using ML and LS estimator. In low SNRs, ML estimator achieves less error compared to the LS estimator.

Association between radioiodine therapy for Graves' hyperthyroidism and thyroid-associated ophthalmopathy.

To investigate the role of radioactive iodine (RAI) in the onset and progression of thyroid-associated ophthalmopathy (TAO). Forty-six Graves' disease patients with mild or no ophthalmopathy were prospectively treated with carbimazole (CBZ) (n = 22) or RAI (n = 24). Treatment effects were evaluated clinically over 12 months, and with orbital MRI-measured extra-ocular muscle (EOM) volumes at baseline and at 6 months. The diagnosis of TAO was based on the clinical activity score (CAS) system. There were 11/22 CBZ and 10/24 RAI patients with active ophthalmopathy at baseline. Despite greater mean TSH levels post-RAI (P = 0.003), there was no increase in the likelihood of developing active ophthalmopathy (OR 0.95; 95% CI 0.56-1.61, P = 0.9) or EOM dysfunction (OR 0.52; 95% CI 0.26-1.06, P = 0.074). The increased mean palpebral aperture post-RAI (P = 0.023) and greater mean proptosis in the CBZ group (P = 0.005) were not confirmed when the absolute values of these measurements were examined. There was no association between the treatment received and MRI-measured EOM volumes. In this study, RAI therapy for Graves' disease did not increase the risk of progression or development of ophthalmopathy in patients with mild or no eye disease at baseline.

Magnetic resonance imaging study in older fragile X premutation male carriers.

Some carriers of a "premutation" allele of the FMR1 gene develop late-onset tremor/ataxia. We conducted a magnetic resonance imaging volumetric study in an unselected sample of eight older male premutation carriers. Volumetric measures, including total brain volume, and the volumes of cerebrum, cerebellum, and cerebral cortex all were significantly reduced in premutation carriers compared with similar data from 21 age-matched normal controls. Total brain and cerebral volumes were significantly related to the number of CGG repeats in the FMR1 gene. Moreover, increased hippocampal volume indicates this premutation may account for both neurodegenerative and neurodevelopmental changes.