An automated flowchart for the Modified Rankin Scale assessment: A multicenter inter-rater agreement analysis

BACKGROUND AND OBJECTIVE

The Modified Rankin Scale (mRS) is a widely adopted scale for assessing stroke recovery. Despite limitations, the mRS has been adopted as primary outcome in most recent clinical acute stroke trials. Designed to be used by multidisciplinary clinical staff, the congruency of this scale is not consistent, which may lead to mistakes in clinical or research application. We aimed to develop and validate an interactive and automated digital tool for assessing the mRS-the iRankin.

METHODS

A panel of five board-certified and mRS-trained vascular neurologists developed an automated flowchart based on current mRS literature. Two international experts were consulted on content and provided feedback on the prototype platform. The platform contained five vignettes and five real video cases, representing mRS grades 0-5. For validation, we invited neurological staff from six comprehensive stroke centers to complete an online assessment. Participants were randomized into two equal groups usual practice versus iRankin. The participants were randomly allocated in pairs for the congruency analysis. Weighted kappa (kw) and proportions were used to describe agreement.

RESULTS

A total of 59 professionals completed the assessment. The kw was dramatically improved among nurses, 0.76 (95% confidence interval (CI) = 0.55-0.97) × 0.30 (0.07-0.67), and among vascular neurologists, 0.87 (0.72-1) × 0.82 (0.66-0.98). In the accuracy analysis, after the standard mRS values for the vignettes and videos were determined by a panel of experts, and considering each correct answer as equivalent to 1 point on a scale of 0-15, it revealed a higher mean of 10.6 (±2.2) in the iRankin group and 8.2 (±2.3) points in the control group (p = 0.02). In an adjusted analysis, the iRankin adoption was independently associated with the score of congruencies between reported and standard scores (beta coefficient = 2.22, 95% CI = 0.64-3.81, p = 0.007).

CONCLUSION

The iRankin adoption led to a substantial or near-perfect agreement in all analyzed professional categories. More trials are needed to generalize our findings. Our user-friendly and free platform is available at https://www.irankinscale.com/.

Artificial intelligence-based automated preprocessing and classification of impacted maxillary canines in panoramic radiographs

OBJECTIVES

Automating the digital workflow for diagnosing impacted canines using panoramic radiographs (PRs) is challenging. This study explored feature extraction, automated cropping, and classification of impacted and nonimpacted canines as a first step.

METHODS

A convolutional neural network with SqueezeNet architecture was first trained to classify two groups of PRs (91with and 91without impacted canines) on the MATLAB programming platform. Based on results, the need to crop the PRs was realized. Next, artificial intelligence (AI) detectors were trained to identify specific landmarks (maxillary central incisors, lateral incisors, canines, bicuspids, nasal area, and the mandibular ramus) on the PRs. Landmarks were then explored to guide cropping of the PRs. Finally, improvements in classification of automatically cropped PRs were studied.

RESULTS

Without cropping, the area under the curve (AUC) of the receiver operating characteristic (ROC) curve for classifying impacted and nonimpacted canine was 84%. Landmark training showed that detectors could correctly identify upper central incisors and the ramus in ∼98% of PRs. The combined use of the mandibular ramus and maxillary central incisors as guides for cropping yielded the best results (∼10% incorrect cropping). When automatically cropped PRs were used, the AUC-ROC improved to 96%.

CONCLUSIONS

AI algorithms can be automated to preprocess PRs and improve the identification of impacted canines.

MATLAB-Based Algorithm and Software for Analysis of Wavy Collagen Fibers

Knowledge of soft tissue fiber structure is necessary for accurate characterization and modeling of their mechanical response. Fiber configuration and structure informs both our understanding of healthy tissue physiology and of pathological processes resulting from diseased states. This study develops an automatic algorithm to simultaneously estimate fiber global orientation, abundance, and waviness in an investigated image. To our best knowledge, this is the first validated algorithm which can reliably separate fiber waviness from its global orientation for considerably wavy fibers. This is much needed feature for biological tissue characterization. The algorithm is based on incremental movement of local regions of interest (ROI) and analyzes two-dimensional images. Pixels belonging to the fiber are identified in the ROI, and ROI movement is determined according to local orientation of fiber within the ROI. The algorithm is validated with artificial images and ten images of porcine trachea containing wavy fibers. In each image, 80-120 fibers were tracked manually to serve as verification. The coefficient of determination R2 between curve lengths and histograms documenting the fiber waviness and global orientation were used as metrics for analysis. Verification-confirmed results were independent of image rotation and degree of fiber waviness, with curve length accuracy demonstrated to be below 1% of fiber curved length. Validation-confirmed median and interquartile range of R2, respectively, were 0.90 and 0.05 for curved length, 0.92 and 0.07 for waviness, and 0.96 and 0.04 for global orientation histograms. Software constructed from the proposed algorithm was able to track one fiber in about 1.1 s using a typical office computer. The proposed algorithm can reliably and accurately estimate fiber waviness, curve length, and global orientation simultaneously, moving beyond the limitations of prior methods.

Real-World Evaluation of an Automated Algorithm to Detect Patients With Potentially Undiagnosed Hypertension Among Patients With Routine Care in Hawai'i

BACKGROUND

This real-world evaluation considers an algorithm designed to detect patients with potentially undiagnosed hypertension, receiving routine care, in a large health system in Hawai'i. It quantifies patients identified as potentially undiagnosed with hypertension; summarizes the individual, clinical, and health system factors associated with undiagnosed hypertension; and examines if the COVID-19 pandemic affected detection.

METHODS AND RESULTS

We analyzed the electronic health records of patients treated across 6 clinics from 2018 to 2021. We calculated total patients with potentially undiagnosed hypertension and compared patients flagged for undiagnosed hypertension to those with diagnosed hypertension and to the full patient panel across individual characteristics, clinical and health system factors (eg, clinic of care), and timing. Modified Poisson regression was used to calculate crude and adjusted risk ratios. Among the eligible patients (N=13 364), 52.6% had been diagnosed with hypertension, 2.7% were flagged as potentially undiagnosed, and 44.6% had no evidence of hypertension. Factors associated with a higher risk of potentially undiagnosed hypertension included individual characteristics (ages 40-84 compared with 18-39 years), clinical (lack of diabetes diagnosis) and health system factors (clinic site and being a Medicaid versus a Medicare beneficiary), and timing (readings obtained after the COVID-19 Stay-At-Home Order in Hawai'i).

CONCLUSIONS

This evaluation provided evidence that a clinical algorithm implemented within a large health system's electronic health records could detect patients in need of follow-up to determine hypertension status, and it identified key individual characteristics, clinical and health system factors, and timing considerations that may contribute to undiagnosed hypertension among patients receiving routine care.

Automated quantification of uveitic keratic precipitates by use of anterior segment optical coherence tomography

BACKGROUND

Evaluation of ocular inflammation via common imaging modalities like optical coherence tomography (OCT) has emphasised cell visualisation, but automated detection of uveitic keratic precipitates (KPs) remains unexplored.

METHODS

Anterior segment (AS)-OCT dense volumes of the corneas of patients with uveitic KPs were collected at three timepoints: with active (T0), clinically improving (T1), and resolved (T2) inflammation. At each visit, visual acuity and clinical grading of the anterior chamber cells were assessed. A bespoke algorithm was used to create an en face rendering of the KPs and to calculate their volume and a ratio of the volume of precipitates over the analysed area. The variation of AS-OCT-derived measurements over time was assessed, and compared with clinical grading.

RESULTS

Twenty eyes from 20 patients (13 females, mean age 39 years) were studied. At T0, the mean volume of the corneal KPs was 0.1727 mm3 , and it significantly reduced to 0.1111 mm3 (p = 0.03) only at T2. The ratio between the volume of the KPs and the corneal area decreased from T0 (0.007) to T1 (0.006; p = 0.2) and T2 (0.004; p = 0.009). There was a statistically significant correlation between the AC cell count and the AS-OCT volume measurements of the KPs at the three time points.

CONCLUSIONS

AS-OCT can image uveitic KPs and through a bespoke algorithm we were able to create an en face rendering allowing us to extrapolate their volume. We found that objective quantification of KPs correlated with inflammatory cell counts in the anterior chamber.

The effectiveness of an automated algorithm as a tool for investigating the cause of anaemia in undiagnosed patients from general practitioners

BACKGROUND

The Dutch guideline algorithm for the analysis of anaemia in patients of general practitioners (GPs) was programmed in a Clinical Decision Support system (CDS-anaemia) to support the process of diagnosing the cause of anaemia in the laboratory. This study investigates the diagnostic yield of the automated anaemia algorithm compared to that of the manual work up by the GP.

METHODS

This retrospective population-based study consisted of 2697 people ≥18 years. Anaemia was defined according to the Dutch College of General Practitioners (DCGP) guideline. Causes of anaemia were based on the DCGP guidelines with the corresponding blood tests. The number of blood tests and causes of anaemia were measured in two separate periods in both the (CDS-anaemia) pilot group and a control group in which routine care was provided.

RESULTS

Patients from GPs supported by CDS-anaemia had higher chances of having more anaemia-related blood tests being performed. This resulted in finding significantly more causes of anaemia in the pilot group compared to the control group with respect to iron deficiency (resp. 31.3% vs 14.5%), possible iron deficiency (resp. 11.4% vs 2.8%), iron deficiency in acute phase (2.6% vs 0.5%), chronic disease/infection/inflammation (23.5% vs 1.9%), vitamin B12 deficiency (4.5% vs 1.9%), possible vitamin B12 deficiency (16.8% vs 8.7%), folate deficiency (3.3% vs 0.9%) and possible bone marrow disorder (3.8% vs 0.0%); p < 0.05.

CONCLUSIONS

This study suggests that an automated-algorithm support can effectively aid in the diagnostic work-up of anaemia in primary care to find more causes of anaemia.

Meibomian Gland Morphology Changes After Cataract Surgery: A Contra-Lateral Eye Study

Purpose: To evaluate the morphology changes of meibomian glands (MGs) after cataract surgery. Setting: Hangzhou Branch of the Eye Hospital of Wenzhou Medical University, Zhejiang, China. Methods: In this contra-lateral eye study, 40 patients received unilateral cataract surgery for age-related cataract. All the patients underwent the evaluation of non-invasive break-up time (NIBUT) and lower tear meniscus height (TMH) before the surgery and 6 months post-operatively. The MGs were evaluated via ImageJ and Meibomian Gland Bio-image Analyzer. MG dropout, length, width, area, gland diameter deformation index (DI), and gland signal index (SI) were recorded. Results: MG length, width, area, DI, and SI were significantly decreased after cataract surgery in the study group (operated eyes, P < 0.001, P = 0.003, P < 0.001, P = 0.001, and P < 0.001, respectively) and showed no significant changes in the control group (non-operated eyes) (all P > 0.05). MG loss increased more in the study group (P = 0.030), and the changes in TMH and NIBUT were not significantly different between the two eyes (both P > 0.05). Conclusion: Cataract surgery aggravated meibomian gland morphology, such as MG loss, MG length, width, area, and SI, and produced no change in NIBUT and TMH at 6 months post-operatively.

An Automated Line-of-Therapy Algorithm for Adults With Metastatic Non-Small Cell Lung Cancer: Validation Study Using Blinded Manual Chart Review

BACKGROUND

Extraction of line-of-therapy (LOT) information from electronic health record and claims data is essential for determining longitudinal changes in systemic anticancer therapy in real-world clinical settings.

OBJECTIVE

The aim of this retrospective cohort analysis is to validate and refine our previously described open-source LOT algorithm by comparing the output of the algorithm with results obtained through blinded manual chart review.

METHODS

We used structured electronic health record data and clinical documents to identify 500 adult patients treated for metastatic non-small cell lung cancer with systemic anticancer therapy from 2011 to mid-2018; we assigned patients to training (n=350) and test (n=150) cohorts, randomly divided proportional to the overall ratio of simple:complex cases (n=254:246). Simple cases were patients who received one LOT and no maintenance therapy; complex cases were patients who received more than one LOT and/or maintenance therapy. Algorithmic changes were performed using the training cohort data, after which the refined algorithm was evaluated against the test cohort.

RESULTS

For simple cases, 16 instances of discordance between the LOT algorithm and chart review prerefinement were reduced to 8 instances postrefinement; in the test cohort, there was no discordance between algorithm and chart review. For complex cases, algorithm refinement reduced the discordance from 68 to 62 instances, with 37 instances in the test cohort. The percentage agreement between LOT algorithm output and chart review for patients who received one LOT was 89% prerefinement, 93% postrefinement, and 93% for the test cohort, whereas the likelihood of precise matching between algorithm output and chart review decreased with an increasing number of unique regimens. Several areas of discordance that arose from differing definitions of LOTs and maintenance therapy could not be objectively resolved because of a lack of precise definitions in the medical literature.

CONCLUSIONS

Our findings identify common sources of discordance between the LOT algorithm and clinician documentation, providing the possibility of targeted algorithm refinement.

A Novel Model Based on Spatial and Morphological Domains to Predict the Origin of Premature Ventricular Contraction

Pace mapping is commonly used to locate the origin of ventricular arrhythmias, especially premature ventricular contraction (PVC). However, this technique relies on clinicians’ ability to rapidly interpret ECG data. To avoid time-consuming interpretation of ECG morphology, some automated algorithms or computational models have been explored to guide the ablation. Inspired by these studies, we propose a novel model based on spatial and morphological domains. The purpose of this study is to assess this model and compare it with three existing models. The data are available from the Experimental Data and Geometric Analysis Repository database in which three in vivo PVC patients are included. To measure the hit rate (A hit occurs when the predicted site is within 15 mm of the target) of different algorithms, 47 target sites are tested. Moreover, to evaluate the efficiency of different models in narrowing down the target range, 54 targets are verified. As a result, the proposed algorithm achieves the most hits (37/47) and fewest misses (9/47), and it narrows down the target range most, from 27.62 ± 3.47 mm to 10.72 ± 9.58 mm among 54 target sites. It is expected to be applied in the real-time prediction of the origin of ventricular activation to guide the clinician toward the target site.

Acetylcholine Regulates Pulmonary Pathology During Viral Infection and Recovery

Introduction

This study was designed to explore the role of acetylcholine (ACh) in pulmonary viral infection and recovery. Inflammatory control is critical to recovery from respiratory viral infection. ACh secreted from non-neuronal sources, including lymphocytes, plays an important, albeit underappreciated, role in regulating immune-mediated inflammation.

Methods

ACh and lymphocyte cholinergic status in the lungs were measured over the course of influenza infection and recovery. The role of ACh was examined by inhibiting ACh synthesis in vivo. Pulmonary inflammation was monitored by Iba1 immunofluorescence, using a novel automated algorithm. Tissue repair was monitored histologically.

Results

Pulmonary ACh remained constant through the early stage of infection and increased during the peak of the acquired immune response. As the concentration of ACh increased, cholinergic lymphocytes appeared in the BAL and lungs. Cholinergic capacity was found primarily in CD4 T cells, but also in B cells and CD8 T cells. The cholinergic CD4⁺ T cells bound to influenza-specific tetramers and were retained in the resident memory regions of the lung up to 2 months after infection. Histologically, cholinergic lymphocytes were found in direct physical contact with activated macrophages throughout the lung. Inflammation was monitored by ionized calcium-binding adapter molecule 1 (Iba1) immunofluorescence, using a novel automated algorithm. When ACh production was inhibited, mice exhibited increased tissue inflammation and delayed recovery. Histologic examination revealed abnormal tissue repair when ACh was limited.

Conclusion

These findings point to a previously unrecognized role for ACh in the transition from active immunity to recovery and pulmonary repair following respiratory viral infection.

The Generalizability of a Medication Administration Discrepancy Detection System: Quantitative Comparative Analysis

Background

As a result of the overwhelming proportion of medication errors occurring each year, there has been an increased focus on developing medication error prevention strategies. Recent advances in electronic health record (EHR) technologies allow institutions the opportunity to identify medication administration error events in real time through computerized algorithms. MED.Safe, a software package comprising medication discrepancy detection algorithms, was developed to meet this need by performing an automated comparison of medication orders to medication administration records (MARs). In order to demonstrate generalizability in other care settings, software such as this must be tested and validated in settings distinct from the development site.

Objective

The purpose of this study is to determine the portability and generalizability of the MED.Safe software at a second site by assessing the performance and fit of the algorithms through comparison of discrepancy rates and other metrics across institutions.

Methods

The MED.Safe software package was executed on medication use data from the implementation site to generate prescribing ratios and discrepancy rates. A retrospective analysis of medication prescribing and documentation patterns was then performed on the results and compared to those from the development site to determine the algorithmic performance and fit. Variance in performance from the development site was further explored and characterized.

Results

Compared to the development site, the implementation site had lower audit/order ratios and higher MAR/(order + audit) ratios. The discrepancy rates on the implementation site were consistently higher than those from the development site. Three drivers for the higher discrepancy rates were alternative clinical workflow using orders with dosing ranges; a data extract, transfer, and load issue causing modified order data to overwrite original order values in the EHRs; and delayed EHR documentation of verbal orders. Opportunities for improvement were identified and applied using a software update, which decreased false-positive discrepancies and improved overall fit.

Conclusions

The execution of MED.Safe at a second site was feasible and effective in the detection of medication administration discrepancies. A comparison of medication ordering, administration, and discrepancy rates identified areas where MED.Safe could be improved through customization. One modification of MED.Safe through deployment of a software update improved the overall algorithmic fit at the implementation site. More flexible customizations to accommodate different clinical practice patterns could improve MED.Safe’s fit at new sites.

An Electronic Search Algorithm for Early Disseminated Intravascular Coagulopathy Diagnosis in the Intensive Care Unit: A Derivation and Validation Study

Aim: We aim to create and validate an electronic search algorithm for accurate detection of disseminated intravascular coagulopathy (DIC) from medical records.

Methods: Patients with DIC in Mayo Clinic’s intensive care units (ICUs) from Jan 1, 2007, to May 4, 2018, were included in the study. An algorithm was developed based on clinical notes and ICD diagnosis codes. A cohort of 50 patients was included with DIC diagnosis, its variations, and no diagnosis of DIC. Then, the next set of 50 patients was used to refine the algorithm. Results were compared with a manual reviewer and the disagreements were resolved by the third reviewer. The same process was repeated with 'revised clinical note search' for the first and second derivation cohort with additional exclusion terms. The obtained sensitivity and specificity were reported. The generated algorithm was applied to another set of 50 patients for validation.

Results: In the first derivation cohort- DIC search by clinical notes and diagnosis codes had 92% sensitivity and 100% specificity. Sensitivity dropped to 71% in the second cohort although specificity remains the same. Therefore, the algorithm was refined to clinical notes search only. The revised search was reapplied to first and second derivation cohorts and results obtained for the first derivation were the same but 91.3% sensitive and 100% specific for the second derivation. The search was locked and applied in the validation cohort with 95.8% sensitivity and 100% specificity, respectively.

Conclusion: The revised clinical note based electronic search algorithm was found to be highly sensitive and specific for DIC during the corresponding ICU duration.

Automated cortical auditory response detection strategy

Objective: This study describes a new automated strategy to determine the detection status of an electrophysiological response.Design: Response, noise and signal-to-noise ratio of the cortical auditory evoked potential (CAEP) were characterised. Detection rules were defined: when to start testing, when to conduct subsequent statistical tests using residual noise as an objective criterion, and when to stop testing.Study sample: Simulations were run to determine optimal parameters on a large combined CAEP data set collected in 45 normal-hearing adults and 17 adults with hearing loss.Results: The proposed strategy to detect CAEPs is fully automated. The first statistical test is conducted when the residual noise level is equal to or smaller than 5.1 µV. The succeeding Hotelling's T² statistical tests are conducted using pre-defined residual noise levels criteria ranging from 5.1 to 1.2 µV. A rule was introduced allowing to stop testing before the maximum number of recorded epochs is reached, depending on a minimum p-value criterion.Conclusion: The proposed framework can be applied to systems which involves detection of electrophysiological responses in biological systems containing background noise. The proposed detection algorithm which optimise sensitivity, specificity, and recording time has the potential to be used in clinical setting.

A Trajectory-Based Method to Explore Reaction Mechanisms

The tsscds method, recently developed in our group, discovers chemical reaction mechanisms with minimal human intervention. It employs accelerated molecular dynamics, spectral graph theory, statistical rate theory and stochastic simulations to uncover chemical reaction paths and to solve the kinetics at the experimental conditions. In the present review, its application to solve mechanistic/kinetics problems in different research areas will be presented. Examples will be given of reactions involved in photodissociation dynamics, mass spectrometry, combustion chemistry and organometallic catalysis. Some planned improvements will also be described.

High-throughput automated scoring of Ki67 in breast cancer tissue microarrays from the Breast Cancer Association Consortium

Automated methods are needed to facilitate high-throughput and reproducible scoring of Ki67 and other markers in breast cancer tissue microarrays (TMAs) in large-scale studies. To address this need, we developed an automated protocol for Ki67 scoring and evaluated its performance in studies from the Breast Cancer Association Consortium. We utilized 166 TMAs containing 16,953 tumour cores representing 9,059 breast cancer cases, from 13 studies, with information on other clinical and pathological characteristics. TMAs were stained for Ki67 using standard immunohistochemical procedures, and scanned and digitized using the Ariol system. An automated algorithm was developed for the scoring of Ki67, and scores were compared to computer assisted visual (CAV) scores in a subset of 15 TMAs in a training set. We also assessed the correlation between automated Ki67 scores and other clinical and pathological characteristics. Overall, we observed good discriminatory accuracy (AUC = 85%) and good agreement (kappa = 0.64) between the automated and CAV scoring methods in the training set. The performance of the automated method varied by TMA (kappa range= 0.37-0.87) and study (kappa range = 0.39-0.69). The automated method performed better in satisfactory cores (kappa = 0.68) than suboptimal (kappa = 0.51) cores (p-value for comparison = 0.005); and among cores with higher total nuclei counted by the machine (4,000-4,500 cells: kappa = 0.78) than those with lower counts (50-500 cells: kappa = 0.41; p-value = 0.010). Among the 9,059 cases in this study, the correlations between automated Ki67 and clinical and pathological characteristics were found to be in the expected directions. Our findings indicate that automated scoring of Ki67 can be an efficient method to obtain good quality data across large numbers of TMAs from multicentre studies. However, robust algorithm development and rigorous pre- and post-analytical quality control procedures are necessary in order to ensure satisfactory performance.

An efficient approach for surveillance of childhood diabetes by type derived from electronic health record data: the SEARCH for Diabetes in Youth Study

OBJECTIVE

To develop an efficient surveillance approach for childhood diabetes by type across 2 large US health care systems, using phenotyping algorithms derived from electronic health record (EHR) data.

MATERIALS AND METHODS

Presumptive diabetes cases <20 years of age from 2 large independent health care systems were identified as those having ≥1 of the 5 indicators in the past 3.5 years, including elevated HbA1c, elevated blood glucose, diabetes-related billing codes, patient problem list, and outpatient anti-diabetic medications. EHRs of all the presumptive cases were manually reviewed, and true diabetes status and diabetes type were determined. Algorithms for identifying diabetes cases overall and classifying diabetes type were either prespecified or derived from classification and regression tree analysis. Surveillance approach was developed based on the best algorithms identified.

RESULTS

We developed a stepwise surveillance approach using billing code-based prespecified algorithms and targeted manual EHR review, which efficiently and accurately ascertained and classified diabetes cases by type, in both health care systems. The sensitivity and positive predictive values in both systems were approximately ≥90% for ascertaining diabetes cases overall and classifying cases with type 1 or type 2 diabetes. About 80% of the cases with "other" type were also correctly classified. This stepwise surveillance approach resulted in a >70% reduction in the number of cases requiring manual validation compared to traditional surveillance methods.

CONCLUSION

EHR data may be used to establish an efficient approach for large-scale surveillance for childhood diabetes by type, although some manual effort is still needed.

Automated analysis of intima-media thickness: analysis and performance of CARES 3.0

OBJECTIVES

In recent years, the use of computer-based techniques has been advocated to improve intima-media thickness (IMT) quantification and its reproducibility. The purpose of this study was to test the diagnostic performance of a new IMT automated algorithm, CARES 3.0, which is a patented class of IMT measurement systems called AtheroEdge (AtheroPoint, LLC, Roseville, CA).

METHODS

From 2 different institutions, we analyzed the carotid arteries of 250 patients. The automated CARES 3.0 algorithm was tested versus 2 other automated algorithms, 1 semiautomated algorithm, and a reader reference to assess the IMT measurements. Bland-Altman analysis, regression analysis, and the Student t test were performed.

RESULTS

CARES 3.0 showed an IMT measurement bias ± SD of -0.022 ± 0.288 mm compared with the expert reader. The average IMT by CARES 3.0 was 0.852 ± 0.248 mm, and that of the reader was 0.872 ± 0.325 mm. In the Bland-Altman plots, the CARES 3.0 IMT measurements showed accurate values, with about 80% of the images having an IMT measurement bias ranging between -50% and +50%. These values were better than those of the previous CARES releases and the semiautomated algorithm. Regression analysis showed that, among all techniques, the best t value was between CARES 3.0 and the reader.

CONCLUSIONS

We have developed an improved fully automated technique for carotid IMT measurement on longitudinal ultrasound images. This new version, called CARES 3.0, consists of a new heuristic for lumen-intima and media-adventitia detection, which showed high accuracy and reproducibility for IMT measurement.