Ameet Patel

Accuracy of dementia diagnosis—a direct comparison between radiologists and a computerized method

There has been recent interest in the application of machine learning techniques to neuroimaging-based diagnosis. These methods promise fully automated, standard PC-based clinical decisions, unbiased by variable radiological expertise. We recently used support vector machines (SVMs) to separate sporadic Alzheimers disease from normal ageing and from fronto-temporal lobar degeneration (FTLD). In this study, we compare the results to those obtained by radiologists. A binary diagnostic classification was made by six radiologists with different levels of experience on the same scans and information that had been previously analysed with SVM. SVMs correctly classified 95% (sensitivity/specificity: 95/95) of sporadic Alzheimers disease and controls into their respective groups. Radiologists correctly classified 65–95% (median 89%; sensitivity/specificity: 88/90) of scans. SVM correctly classified another set of sporadic Alzheimers disease in 93% (sensitivity/specificity: 100/86) of cases, whereas radiologists ranged between 80% and 90% (median 83%; sensitivity/specificity: 80/85). SVMs were better at separating patients with sporadic Alzheimers disease from those with FTLD (SVM 89%; sensitivity/specificity: 83/95; compared to radiological range from 63% to 83%; median 71%; sensitivity/specificity: 64/76). Radiologists were always accurate when they reported a high degree of diagnostic confidence. The results show that well-trained neuroradiologists classify typical Alzheimers disease-associated scans comparable to SVMs. However, SVMs require no expert knowledge and trained SVMs can readily be exchanged between centres for use in diagnostic classification. These results are encouraging and indicate a role for computerized diagnostic methods in clinical practice.

Localization of a Rapid CSF Leak with Digital Subtraction Myelography

SUMMARY: A 53-year-old woman with superficial siderosis underwent spinal MR imaging, which demonstrated a large cervicothoracic epidural fluid collection compatible with a CSF leak. Conventional and dynamic CT myelography failed to localize the dural tear because of rapid equilibration of myelographic contrast between the thecal sac and the extradural collection. The superior temporal resolution of digital subtraction myelography precisely localized the CSF leak preoperatively and led to the successful surgical correction of the dural tear.

Radiation Dose Reduction in Paranasal Sinus CT Using Model-Based Iterative Reconstruction

BACKGROUND AND PURPOSE: CT performed with Veo model-based iterative reconstruction has shown the potential for radiation-dose reduction. This study sought to determine whether Veo could reduce noise and improve the image quality of low-dose sinus CT. MATERIALS AND METHODS: Twenty patients consented to participate and underwent low- and standard-dose sinus CT on the same day. Standard-dose CT was created with filtered back-projection (120 kV[peak], 210 mA, 0.4-second rotation, and 0.531 pitch). For low-dose CT, mA was decreased to 20 (the remaining parameters were unchanged), and images were generated with filtered back-projection and Veo. Standard- and low-dose datasets were reconstructed by using bone and soft-tissue algorithms, while the low-dose Veo reconstruction only had a standard kernel. Two blinded neuroradiologists independently evaluated the image quality of multiple osseous and soft-tissue craniofacial structures. Image noise was measured by using multiple regions of interest. RESULTS: Eight women and 12 men (mean age, 63.3 years) participated. Volume CT dose indices were 2.9 mGy (low dose) and 31.6 mGy (standard dose), and mean dose-length products were 37.4 mGy-cm (low dose) and 406.1 mGy-cm (standard dose). Of all the imaging series, low-dose Veo demonstrated the least noise (P < .001). Compared with filtered back-projection low-dose CT using soft-tissue and bone algorithms, Veo had the best soft-tissue image quality but the poorest bone image quality (P < .001). CONCLUSIONS: Veo significantly reduces noise in low-dose sinus CT. Although this reduction improves soft-tissue evaluation, thin bone becomes less distinct.

CT Interpretation in a Telestroke Network: Agreement Among a Spoke Radiologist, Hub Vascular Neurologist, and Hub Neuroradiologist

Background and Purpose— The American Stroke Association guidelines emphasized the need for further high-quality studies that assess agreement by radiologists and nonradiologists engaged in emergency telestroke assessments and decision-making. Therefore, the objective of this study was to determine the level of agreement of baseline brain CT scan interpretations of patients with acute stroke presenting to telestroke spoke hospitals between central reading committee neuroradiologists and each of 2 groups, spoke hospital radiologists and hub hospital vascular neurologists (telestrokologists). Methods— The Stroke Team Remote Evaluation Using a Digital Observation Camera Arizona trial was a prospective, urban single-hub, rural 2-spoke, randomized, blinded, controlled trial of a 2-way, site-independent, audiovisual telemedicine and teleradiology system designed for remote evaluation of adult patients with acute stroke versus telephone consultation to assess eligibility for treatment with intravenous thrombolysis. In the telemedicine arm, the subjects’ CT scans were interpreted by the hub telestrokologist and in the telephone arm by the spoke radiologist. All subjects’ CT scans were subsequently interpreted centrally, independently, and blindly by 2 hub neuroradiologists. The primary CT outcome was determination of a CT-based contraindication to thrombolytic treatment. Kappa statistics and exact agreement rates were used to analyze interobserver agreement. Results— Fifty-four subjects underwent random assignment. The overall agreement for the presence of radiological contraindications to thrombolysis was excellent (0.91) and did not differ substantially between the hub telestrokologist to neuroradiologist and spoke radiologist to neuroradiologist (0.92 and 0.89, respectively). Conclusions— In the context of a telestroke network designed to assess patients with acute stroke syndromes, agreement over the presence or absence of radiological contraindications to thrombolysis was excellent whether the comparisons were between a telestrokologist and neuroradiologist or between spoke radiologist and neuroradiologist. Clinical Trial Registration— URL: http://www.clinicaltrials.gov. Unique identifier: NCT00623350.

Conscious vs. subconscious perception, as a function of radiological expertise

Radiological images constitute a special class of images that are captured (or computed) for a specific purpose (i.e. diagnosis) and their “correct” interpretation is vitally important. However, because they are not “natural” images, radiologists must be trained to visually interpret them. This training involves perceptual learning that is gradually acquired over an extended period of exposure to radiological images. This implicit (subconscious) knowledge is difficult to pass along explicitly (i.e. verbally) to less experienced radiologists. Multimedia technology has the potential to facilitate perceptual learning in new radiologists. However, it is important to have an objective and quantitative method for evaluating the progress of trainees using this approach. This paper proposes an eye-tracker-based metric for determining the level of expertise of a radiologist in training, based on where he/she lies along a scale based on the visual scanning behavior of radiologists, ranging from novice to expert.

Efficient atypicality detection in chest radiographs

Expert radiologists are able to quickly detect atypical features in chest radiographs because they have developed a sense of what textures and contours are typical for each anatomic region by viewing a large set of “normal” chest radiographs. Our previous work modeled this type of learning with a transductive One-Nearest-Neighbor (1NN) method that was effective for identifying atypical regions in chest radiographs. However, the need to compute distances between the feature vectors extracted from a test image and a very large archive of feature vectors (previously extracted from corresponding anatomical locations in a large set of “normal” chest radiographs) made the 1NN method very computationally intensive. This paper uses an instance selection mechanism based on an Extended Fuzzy C-Means (EFCM) clustering algorithm to reduce the magnitude of this computation. Our results (based on a large set of real-world chest radiographs obtained from Mayo Clinic) indicate that EFCM can substantially reduce the computational cost of the 1NN method, without a substantial drop in the accuracy of its atypicality estimates.

Toward the detection of abnormal chest radiographs the way radiologists do it

Computer Aided Detection (CADe) and Computer Aided Diagnosis (CADx) are relatively recent areas of research that attempt to employ feature extraction, pattern recognition, and machine learning algorithms to aid radiologists in detecting and diagnosing abnormalities in medical images. However, these computational methods are based on the assumption that there are distinct classes of abnormalities, and that each class has some distinguishing features that set it apart from other classes. However, abnormalities in chest radiographs tend to be very heterogeneous. The literature suggests that thoracic (chest) radiologists develop their ability to detect abnormalities by developing a sense of what is normal, so that anything that is abnormal attracts their attention. This paper discusses an approach to CADe that is based on a technique called anomaly detection (which aims to detect outliers in data sets) for the purpose of detecting atypical regions in chest radiographs. However, in order to apply anomaly detection to chest radiographs, it is necessary to develop a basis for extracting features from corresponding anatomical locations in different chest radiographs. This paper proposes a method for doing this, and describes how it can be used to support CADe.

Pantopaque contrast mimicking intracanalicular vestibular schwannoma

Pantopaque (iophendylate) is an oily contrast medium historically used during spine imaging. Due to its persistence in the subarachnoid space and the potential to lead to severe arachnoiditis, it is no longer used today. We present a 40‐year‐old male with new‐onset headaches, imbalance, and vertigo. Brain magnetic resonance imaging revealed a 2‐mm T1 ‐hyperintense intracanalicular lesion. Numerous hyperdense foci were scattered throughout the subarachnoid space on computed tomography. Further history revealed the patient received Pantopaque 30 years prior, after sustaining spinal trauma. Remnant Pantopaque contrast is an important differential when evaluating a patient with a suspected intracranial tumor in order to avoid unwarranted surgical intervention. Laryngoscope, 127:1916–1919, 2017

CT Interpretation in a Telestroke Network

Background and Purpose— The American Stroke Association guidelines emphasized the need for further high-quality studies that assess agreement by radiologists and nonradiologists engaged in emergency telestroke assessments and decision-making. Therefore, the objective of this study was to determine the level of agreement of baseline brain CT scan interpretations of patients with acute stroke presenting to telestroke spoke hospitals between central reading committee neuroradiologists and each of 2 groups, spoke hospital radiologists and hub hospital vascular neurologists (telestrokologists). Methods— The Stroke Team Remote Evaluation Using a Digital Observation Camera Arizona trial was a prospective, urban single-hub, rural 2-spoke, randomized, blinded, controlled trial of a 2-way, site-independent, audiovisual telemedicine and teleradiology system designed for remote evaluation of adult patients with acute stroke versus telephone consultation to assess eligibility for treatment with intravenous thrombolysis. In the telemedicine arm, the subjects’ CT scans were interpreted by the hub telestrokologist and in the telephone arm by the spoke radiologist. All subjects’ CT scans were subsequently interpreted centrally, independently, and blindly by 2 hub neuroradiologists. The primary CT outcome was determination of a CT-based contraindication to thrombolytic treatment. Kappa statistics and exact agreement rates were used to analyze interobserver agreement. Results— Fifty-four subjects underwent random assignment. The overall agreement for the presence of radiological contraindications to thrombolysis was excellent (0.91) and did not differ substantially between the hub telestrokologist to neuroradiologist and spoke radiologist to neuroradiologist (0.92 and 0.89, respectively). Conclusions— In the context of a telestroke network designed to assess patients with acute stroke syndromes, agreement over the presence or absence of radiological contraindications to thrombolysis was excellent whether the comparisons were between a telestrokologist and neuroradiologist or between spoke radiologist and neuroradiologist. Clinical Trial Registration— URL: http://www.clinicaltrials.gov. Unique identifier: NCT00623350.

Computed Tomography Interpretation in a Telestroke Network: Agreement between Spoke Radiologist, Hub Vascular Neurologist, and Hub Neuroradiologist

Background and Purpose— The American Stroke Association guidelines emphasized the need for further high-quality studies that assess agreement by radiologists and nonradiologists engaged in emergency telestroke assessments and decision-making. Therefore, the objective of this study was to determine the level of agreement of baseline brain CT scan interpretations of patients with acute stroke presenting to telestroke spoke hospitals between central reading committee neuroradiologists and each of 2 groups, spoke hospital radiologists and hub hospital vascular neurologists (telestrokologists). Methods— The Stroke Team Remote Evaluation Using a Digital Observation Camera Arizona trial was a prospective, urban single-hub, rural 2-spoke, randomized, blinded, controlled trial of a 2-way, site-independent, audiovisual telemedicine and teleradiology system designed for remote evaluation of adult patients with acute stroke versus telephone consultation to assess eligibility for treatment with intravenous thrombolysis. In the telemedicine arm, the subjects’ CT scans were interpreted by the hub telestrokologist and in the telephone arm by the spoke radiologist. All subjects’ CT scans were subsequently interpreted centrally, independently, and blindly by 2 hub neuroradiologists. The primary CT outcome was determination of a CT-based contraindication to thrombolytic treatment. Kappa statistics and exact agreement rates were used to analyze interobserver agreement. Results— Fifty-four subjects underwent random assignment. The overall agreement for the presence of radiological contraindications to thrombolysis was excellent (0.91) and did not differ substantially between the hub telestrokologist to neuroradiologist and spoke radiologist to neuroradiologist (0.92 and 0.89, respectively). Conclusions— In the context of a telestroke network designed to assess patients with acute stroke syndromes, agreement over the presence or absence of radiological contraindications to thrombolysis was excellent whether the comparisons were between a telestrokologist and neuroradiologist or between spoke radiologist and neuroradiologist. Clinical Trial Registration— URL: http://www.clinicaltrials.gov. Unique identifier: NCT00623350.