Shariq Mohammed
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Shariq Mohammed has a total of 90 co-authors. Here are the top co-authors by number of shared publications:
- Arvind Rao
University of Michigan
5 shared publications - Dipak. K. Dey
Department of Statistics, University of Connecticut, Storrs, USA
4 shared publications - Aritra Halder
Department of Biostatistics & Epidemiology, Drexel University
2 shared publications Arvind Rao
Department of Biostatistics, University of Michigan, 1415 Washington Heights, Ann Arbor, MI 48103, United States; Department of Computational Medicine and Bioinformatics, University of Michigan, 100 Washtenaw Avenue, Ann Arbor, MI 48109, United States; Department of Radiation Oncology, University of Michigan, 1500 E Medical Center Dr, Ann Arbor, MI 48109, United States
2 shared publicationsChunyu Liu
Department of Biostatistics, Boston University School of Public Health, Boston, MA
2 shared publicationsDipak K. Dey
2 shared publications
Latest Publications
- Open Access
Network representations of digital clock drawings for cognitive impairment detection
INTRODUCTION: The digital clock drawing test (dCDT) is a cognitive screening tool capturing high-resolution pen-movement data. Existing approaches for classifying cognitive impairment from dCDTs rely on selected features or image-based methods. We propose representing dCDTs as structural networks to classify cognitive impairment and characterize spatial organization in clock drawings. METHODS: Weighted networks were constructed from 3,362 Framingham Heart Study participants, with and without cognitive impairment. Network representations were generated separately for command and copy conditions, with nodes representing pre-defined clock components (digits, hands, clock face) and edges capturing pairwise spatial relationships between components. Random forest models and graph convolutional networks were trained to predict cognitive impairment from the networks. Models were benchmarked against demographics-only and established summary feature baselines. RESULTS: Network-based models achieved AUC values comparable to the established summary feature baseline across command-only and copy-only conditions. Differences in spatial organization between the cognitively intact and impaired groups were greatest in the upper-right clock region, involving digits 1 and 2, the clock hands, and the surrounding clock face. Differences between the command and copy networks were larger among cognitively impaired participants. DISCUSSION: The proposed network representations provide an objective, scalable, and flexible framework for representing clock drawings for use in classification models. dCDT networks matched the predictive performance of established summary feature-based approaches and identified clock regions where spatial organization differed by cognitive status. After external validation, this approach has the potential to be incorporated within a clinical workflow to support automated and interpretable cognitive screening.
Open AccessNetwork representations of digital clock drawings for cognitive impairment detection
INTRODUCTION: The digital clock drawing test (dCDT) is a cognitive screening tool capturing high-resolution pen-movement data. Existing approaches for classifying cognitive impairment from dCDTs rely on selected features or image-based methods. We propose representing dCDTs as structural networks to classify cognitive impairment and characterize spatial organization in clock drawings. METHODS: Weighted networks were constructed from 3,362 Framingham Heart Study participants, with and without cognitive impairment. Network representations were generated separately for command and copy conditions, with nodes representing pre-defined clock components (digits, hands, clock face) and edges capturing pairwise spatial relationships between components. Random forest models and graph convolutional networks were trained to predict cognitive impairment from the networks. Models were benchmarked against demographics-only and established summary feature baselines. RESULTS: Network-based models achieved AUC values comparable to the established summary feature baseline across command-only and copy-only conditions. Differences in spatial organization between the cognitively intact and impaired groups were greatest in the upper-right clock region, involving digits 1 and 2, the clock hands, and the surrounding clock face. Differences between the command and copy networks were larger among cognitively impaired participants. DISCUSSION: The proposed network representations provide an objective, scalable, and flexible framework for representing clock drawings for use in classification models. dCDT networks matched the predictive performance of established summary feature-based approaches and identified clock regions where spatial organization differed by cognitive status. After external validation, this approach has the potential to be incorporated within a clinical workflow to support automated and interpretable cognitive screening.
- Open Access
Predicting cognitive impairment using novel functional features of spatial proximity and circularity in the digital clock drawing test
The digital clock drawing test (dCDT) is a cognitive screening tool employing a digital pen. While many studies rely on summary statistics of dCDT features to predict cognitive outcomes, these approaches often involve subjective decisions such as feature selection and imputation. In this study, we introduce novel dCDT features, expressed as mathematical functions, and compare them to commonly used summary features. We included dCDTs from 3,415 participants from the Framingham Heart Study. Random forest models with five-fold cross-validation were trained to distinguish participants with mild cognitive impairment or dementia from cognitively intact participants. When combined with established time-based features, functional features related to spatial proximity and circularity demonstrated predictive power comparable to commonly used summary features. Our findings highlight the potential of integrating functional features to detect subtle motions and behaviors in digital cognitive assessments, offering new tools that may enhance diagnostic accuracy and support early detection strategies.
Open AccessPredicting cognitive impairment using novel functional features of spatial proximity and circularity in the digital clock drawing test
The digital clock drawing test (dCDT) is a cognitive screening tool employing a digital pen. While many studies rely on summary statistics of dCDT features to predict cognitive outcomes, these approaches often involve subjective decisions such as feature selection and imputation. In this study, we introduce novel dCDT features, expressed as mathematical functions, and compare them to commonly used summary features. We included dCDTs from 3,415 participants from the Framingham Heart Study. Random forest models with five-fold cross-validation were trained to distinguish participants with mild cognitive impairment or dementia from cognitively intact participants. When combined with established time-based features, functional features related to spatial proximity and circularity demonstrated predictive power comparable to commonly used summary features. Our findings highlight the potential of integrating functional features to detect subtle motions and behaviors in digital cognitive assessments, offering new tools that may enhance diagnostic accuracy and support early detection strategies.
- Open Access
- 29 April 2025
- Published by Wolters Kluwer Health in Critical Care Explorations
- Vol. 7 (5) , e1257
Early Pupil Abnormality Frequency Predicts Poor Outcomes and Enhances International Mission for Prognosis and Analysis of Clinical Trials in Traumatic Brain Injury (IMPACT) Model Prognostication in Traumatic Brain Injury
IMPORTANCE: In patients with traumatic brain injury (TBI), baseline pupillary assessment is routine; however, the occurrence rate and clinical significance of pupil abnormalities over the early course of hospitalization remain poorly characterized. OBJECTIVES: To determine whether the occurrence and frequency of pupil abnormalities within the first 72 hours of ICU admission are associated with unfavorable discharge outcomes and to assess whether incorporating this frequency improves the performance of an established prognostic model. DESIGN, SETTING, AND PARTICIPANTS: This was a retrospective observational study of adults admitted with a primary diagnosis of TBI to a single tertiary care ICU between 2018 and 2022. Inclusion criteria included at least three quantitative pupillometry assessments within the first 72 hours. MAIN OUTCOMES AND MEASURES: Quantitative pupillometry was used to calculate the Neurological Pupil index (NPi) at each assessment. Abnormalities were defined as NPi less than 3 in either eye, NPi asymmetry greater than or equal to 0.7, or pupil size asymmetry greater than or equal to 1 mm. The primary outcome was unfavorable discharge disposition (death, hospice, or long-term care). Multivariable logistic regression was used to evaluate the association between pupil abnormality frequency and outcomes, and model performance was compared using goodness-of-fit tests with and without pupil frequency added to the International Mission for Prognosis and Analysis of Clinical Trials in TBI (IMPACT) model. RESULTS: Among 131 patients (median age, 59 yr; 30% women), 35% had an unfavorable discharge disposition. Pupil abnormalities occurred in 60% of mild, 61% of moderate, and 88% of severe TBI patients. For each 1% increase in the frequency of pupil abnormalities over 72 hours, the odds of unfavorable discharge increased by 3% (odds ratio, 1.03; 95% CI, 1.01–1.05). Adding pupil abnormality frequency to the IMPACT model improved its goodness-of-fit (χ2 = 5.24; p = 0.02). CONCLUSIONS AND RELEVANCE: Pupil abnormalities are common across TBI severities, particularly in severe cases. A higher frequency of abnormal pupil measurements within the first 72 hours is associated with unfavorable outcomes and significantly enhances the predictive performance of established TBI prognostic models. Serial quantitative pupillometry may offer clinically valuable, dynamic prognostic information in the acute care setting.
- 29 April 2025
- Published by Wolters Kluwer Health in Critical Care Explorations
- Vol. 7 (5) , e1257
Open AccessEarly Pupil Abnormality Frequency Predicts Poor Outcomes and Enhances International Mission for Prognosis and Analysis of Clinical Trials in Traumatic Brain Injury (IMPACT) Model Prognostication in Traumatic Brain Injury
IMPORTANCE: In patients with traumatic brain injury (TBI), baseline pupillary assessment is routine; however, the occurrence rate and clinical significance of pupil abnormalities over the early course of hospitalization remain poorly characterized. OBJECTIVES: To determine whether the occurrence and frequency of pupil abnormalities within the first 72 hours of ICU admission are associated with unfavorable discharge outcomes and to assess whether incorporating this frequency improves the performance of an established prognostic model. DESIGN, SETTING, AND PARTICIPANTS: This was a retrospective observational study of adults admitted with a primary diagnosis of TBI to a single tertiary care ICU between 2018 and 2022. Inclusion criteria included at least three quantitative pupillometry assessments within the first 72 hours. MAIN OUTCOMES AND MEASURES: Quantitative pupillometry was used to calculate the Neurological Pupil index (NPi) at each assessment. Abnormalities were defined as NPi less than 3 in either eye, NPi asymmetry greater than or equal to 0.7, or pupil size asymmetry greater than or equal to 1 mm. The primary outcome was unfavorable discharge disposition (death, hospice, or long-term care). Multivariable logistic regression was used to evaluate the association between pupil abnormality frequency and outcomes, and model performance was compared using goodness-of-fit tests with and without pupil frequency added to the International Mission for Prognosis and Analysis of Clinical Trials in TBI (IMPACT) model. RESULTS: Among 131 patients (median age, 59 yr; 30% women), 35% had an unfavorable discharge disposition. Pupil abnormalities occurred in 60% of mild, 61% of moderate, and 88% of severe TBI patients. For each 1% increase in the frequency of pupil abnormalities over 72 hours, the odds of unfavorable discharge increased by 3% (odds ratio, 1.03; 95% CI, 1.01–1.05). Adding pupil abnormality frequency to the IMPACT model improved its goodness-of-fit (χ2 = 5.24; p = 0.02). CONCLUSIONS AND RELEVANCE: Pupil abnormalities are common across TBI severities, particularly in severe cases. A higher frequency of abnormal pupil measurements within the first 72 hours is associated with unfavorable outcomes and significantly enhances the predictive performance of established TBI prognostic models. Serial quantitative pupillometry may offer clinically valuable, dynamic prognostic information in the acute care setting.
- 20 March 2025
- Published by Wolters Kluwer Health in Journal of Bone and Joint Surgery
- Vol. 107 (9) , 976-984
Association Between Tourniquet Use and Patient-Reported Outcomes Following Total Knee Arthroplasty
Background: Total knee arthroplasty (TKA) is one of the most commonly performed elective procedures in North America. While advancements have been made in patient optimization, surgical technique, and implant design, tourniquet use remains a contentious issue as it relates to patient outcomes and postoperative experience. Methods: As part of the PEPPER trial, we identified 5,684 patients who underwent primary TKA, of whom 4,866 (85.6%) underwent surgery with a tourniquet (the YT group) and 818 (14.4%) underwent surgery without a tourniquet (the NT group). The cohort was predominantly female (60.8%), White (77%), and of an ethnicity other than Hispanic or Latino (96.8%). The mean age of the patients was 64.6 ± 9.2 years. The primary outcomes were the Knee injury and Osteoarthritis Outcome Score, Joint Replacement (KOOS JR); Patient-Reported Outcomes Measurement Information System Physical Health Summary (PROMIS-PH10); and numeric pain rating scale (NPRS), which were captured preoperatively and at 1, 3, and 6 months postoperatively. The secondary outcomes were length of stay, discharge disposition, analgesic consumption, and postoperative complications. Multivariable analysis was performed to assess the associations between tourniquet use and patient-reported outcome measures (PROMs) following TKA. Results: The percentages of patients achieving the minimal clinically important difference (MCID) for the KOOS JR were significantly different at 1 month only (YT, 55.4%; NT, 47.9%). This difference disappeared at 3 and 6 months. There was no difference between the YT and NT groups in terms of the percentage of patients achieving the MCID for the PROMIS-PH10 or NPRS at any time point. There were no differences between the YT and NT groups at any time point with respect to the KOOS JR, PROMIS-PH10, and NPRS. There were no differences in opioid consumption, operative time, length of stay, wound-related complications, or readmissions postoperatively. Conclusions: Tourniquet use was associated with more patients achieving the MCID for the KOOS JR at 1 month compared with no tourniquet use. This difference disappeared at 3 and 6 months. At 1, 3, and 6 months, there were no differences in opioid consumption, health-care utilization, or complications between patients undergoing TKA with a tourniquet versus without a tourniquet. Tourniquet use did not have a clinically meaningful impact on PROMs in the multivariable analysis. Arthroplasty surgeons may use these data during preoperative discussions with patients regarding tourniquet use as it relates to the surgeon’s preference and how it could influence postoperative function. Level of Evidence: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.
- 20 March 2025
- Published by Wolters Kluwer Health in Journal of Bone and Joint Surgery
- Vol. 107 (9) , 976-984
Association Between Tourniquet Use and Patient-Reported Outcomes Following Total Knee Arthroplasty
Background: Total knee arthroplasty (TKA) is one of the most commonly performed elective procedures in North America. While advancements have been made in patient optimization, surgical technique, and implant design, tourniquet use remains a contentious issue as it relates to patient outcomes and postoperative experience. Methods: As part of the PEPPER trial, we identified 5,684 patients who underwent primary TKA, of whom 4,866 (85.6%) underwent surgery with a tourniquet (the YT group) and 818 (14.4%) underwent surgery without a tourniquet (the NT group). The cohort was predominantly female (60.8%), White (77%), and of an ethnicity other than Hispanic or Latino (96.8%). The mean age of the patients was 64.6 ± 9.2 years. The primary outcomes were the Knee injury and Osteoarthritis Outcome Score, Joint Replacement (KOOS JR); Patient-Reported Outcomes Measurement Information System Physical Health Summary (PROMIS-PH10); and numeric pain rating scale (NPRS), which were captured preoperatively and at 1, 3, and 6 months postoperatively. The secondary outcomes were length of stay, discharge disposition, analgesic consumption, and postoperative complications. Multivariable analysis was performed to assess the associations between tourniquet use and patient-reported outcome measures (PROMs) following TKA. Results: The percentages of patients achieving the minimal clinically important difference (MCID) for the KOOS JR were significantly different at 1 month only (YT, 55.4%; NT, 47.9%). This difference disappeared at 3 and 6 months. There was no difference between the YT and NT groups in terms of the percentage of patients achieving the MCID for the PROMIS-PH10 or NPRS at any time point. There were no differences between the YT and NT groups at any time point with respect to the KOOS JR, PROMIS-PH10, and NPRS. There were no differences in opioid consumption, operative time, length of stay, wound-related complications, or readmissions postoperatively. Conclusions: Tourniquet use was associated with more patients achieving the MCID for the KOOS JR at 1 month compared with no tourniquet use. This difference disappeared at 3 and 6 months. At 1, 3, and 6 months, there were no differences in opioid consumption, health-care utilization, or complications between patients undergoing TKA with a tourniquet versus without a tourniquet. Tourniquet use did not have a clinically meaningful impact on PROMs in the multivariable analysis. Arthroplasty surgeons may use these data during preoperative discussions with patients regarding tourniquet use as it relates to the surgeon’s preference and how it could influence postoperative function. Level of Evidence: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.
- Open Access
- 22 April 2024
- Published by Springer Nature in Scientific Reports
- Vol. 14 (1) , 1-12
Modeling health and well-being measures using ZIP code spatial neighborhood patterns
Individual-level assessment of health and well-being permits analysis of community well-being and health risk evaluations across several dimensions of health. It also enables comparison and rankings of reported health and well-being for large geographical areas such as states, metropolitan areas, and counties. However, there is large variation in reported well-being within such large spatial units underscoring the importance of analyzing well-being at more granular levels, such as ZIP codes. In this paper, we address this problem by modeling well-being data to generate ZIP code tabulation area (ZCTA)-level rankings through spatially informed statistical modeling. We build regression models for individual-level overall well-being index and scores from five subscales (Physical, Financial, Social, Community, Purpose) using individual-level demographic characteristics as predictors while including a ZCTA-level spatial effect. The ZCTA neighborhood information is incorporated by using a graph Laplacian matrix; this enables estimation of the effect of a ZCTA on well-being using individual-level data from that ZCTA as well as by borrowing information from neighboring ZCTAs. We deploy our model on well-being data for the U.S. states of Massachusetts and Georgia. We find that our model can capture the effects of demographic features while also offering spatial effect estimates for all ZCTAs, including ones with no observations, under certain conditions. These spatial effect estimates provide community health and well-being rankings of ZCTAs, and our method can be deployed more generally to model other outcomes that are spatially dependent as well as data from other states or groups of states.
- 22 April 2024
- Published by Springer Nature in Scientific Reports
- Vol. 14 (1) , 1-12
Open AccessModeling health and well-being measures using ZIP code spatial neighborhood patterns
Individual-level assessment of health and well-being permits analysis of community well-being and health risk evaluations across several dimensions of health. It also enables comparison and rankings of reported health and well-being for large geographical areas such as states, metropolitan areas, and counties. However, there is large variation in reported well-being within such large spatial units underscoring the importance of analyzing well-being at more granular levels, such as ZIP codes. In this paper, we address this problem by modeling well-being data to generate ZIP code tabulation area (ZCTA)-level rankings through spatially informed statistical modeling. We build regression models for individual-level overall well-being index and scores from five subscales (Physical, Financial, Social, Community, Purpose) using individual-level demographic characteristics as predictors while including a ZCTA-level spatial effect. The ZCTA neighborhood information is incorporated by using a graph Laplacian matrix; this enables estimation of the effect of a ZCTA on well-being using individual-level data from that ZCTA as well as by borrowing information from neighboring ZCTAs. We deploy our model on well-being data for the U.S. states of Massachusetts and Georgia. We find that our model can capture the effects of demographic features while also offering spatial effect estimates for all ZCTAs, including ones with no observations, under certain conditions. These spatial effect estimates provide community health and well-being rankings of ZCTAs, and our method can be deployed more generally to model other outcomes that are spatially dependent as well as data from other states or groups of states.
- Open Access
- 12 September 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 90, 102964
Tumor radiogenomics in gliomas with Bayesian layered variable selection
- 12 September 2023
- Published by Elsevier in Medical Image Analysis
- Vol. 90, 102964
Open AccessTumor radiogenomics in gliomas with Bayesian layered variable selection
- Open Access
- 19 June 2023
- Published by New England Statistical Society in The New England Journal of Statistics in Data Science
- p. 187-199
Bayesian Variable Selection in Double Generalized Linear Tweedie Spatial Process Models
Publisher: New England Statistical Society, Journal: The New England Journal of Statistics in Data Science, Title: Bayesian Variable Selection in Double Generalized Linear Tweedie Spatial Process Models, Authors: Aritra Halder, Shariq Mohammed, Dipak K. Dey , Double generalized linear models provide a flexible framework for modeling data by allowing the mean and the dispersion to vary across observations. Common members of the exponential dispersion family including the Gaussian, Poisson, compound Poisson-gamma (CP-g), Gamma and inverse-Gaussian are known to admit such models. The lack of their use can be attributed to ambiguities that exist in model specification under a large number of covariates and complications that arise when data display complex spatial dependence. In this work we consider a hierarchical specification for the CP-g model with a spatial random effect. The spatial effect is targeted at performing uncertainty quantification by modeling dependence within the data arising from location based indexing of the response. We focus on a Gaussian process specification for the spatial effect. Simultaneously, we tackle the problem of model specification for such models using Bayesian variable selection. It is effected through a continuous spike and slab prior on the model parameters, specifically the fixed effects. The novelty of our contribution lies in the Bayesian frameworks developed for such models. We perform various synthetic experiments to showcase the accuracy of our frameworks. They are then applied to analyze automobile insurance premiums in Connecticut, for the year of 2008.
- 19 June 2023
- Published by New England Statistical Society in The New England Journal of Statistics in Data Science
- p. 187-199
Open AccessBayesian Variable Selection in Double Generalized Linear Tweedie Spatial Process Models
Publisher: New England Statistical Society, Journal: The New England Journal of Statistics in Data Science, Title: Bayesian Variable Selection in Double Generalized Linear Tweedie Spatial Process Models, Authors: Aritra Halder, Shariq Mohammed, Dipak K. Dey , Double generalized linear models provide a flexible framework for modeling data by allowing the mean and the dispersion to vary across observations. Common members of the exponential dispersion family including the Gaussian, Poisson, compound Poisson-gamma (CP-g), Gamma and inverse-Gaussian are known to admit such models. The lack of their use can be attributed to ambiguities that exist in model specification under a large number of covariates and complications that arise when data display complex spatial dependence. In this work we consider a hierarchical specification for the CP-g model with a spatial random effect. The spatial effect is targeted at performing uncertainty quantification by modeling dependence within the data arising from location based indexing of the response. We focus on a Gaussian process specification for the spatial effect. Simultaneously, we tackle the problem of model specification for such models using Bayesian variable selection. It is effected through a continuous spike and slab prior on the model parameters, specifically the fixed effects. The novelty of our contribution lies in the Bayesian frameworks developed for such models. We perform various synthetic experiments to showcase the accuracy of our frameworks. They are then applied to analyze automobile insurance premiums in Connecticut, for the year of 2008.
- Open Access
- 16 August 2022
- Published by Oxford University Press (OUP) in Biometrics
- Vol. 79 (3) , 1801-1813
Integrative Bayesian Models Using Post-Selective Inference: A Case Study in Radiogenomics
Integrative analyses based on statistically relevant associations between genomics and a wealth of intermediary phenotypes (such as imaging) provide vital insights into their clinical relevance in terms of the disease mechanisms. Estimates for uncertainty in the resulting integrative models are however unreliable unless inference accounts for the selection of these associations with accuracy. In this paper, we develop selection-aware Bayesian methods, which (1) counteract the impact of model selection bias through a “selection-aware posterior” in a flexible class of integrative Bayesian models post a selection of promising variables via ℓ1-regularized algorithms; (2) strike an inevitable trade-off between the quality of model selection and inferential power when the same data set is used for both selection and uncertainty estimation. Central to our methodological development, a carefully constructed conditional likelihood function deployed with a reparameterization mapping provides tractable updates when gradient-based Markov chain Monte Carlo (MCMC) sampling is used for estimating uncertainties from the selection-aware posterior. Applying our methods to a radiogenomic analysis, we successfully recover several important gene pathways and estimate uncertainties for their associations with patient survival times.
- 16 August 2022
- Published by Oxford University Press (OUP) in Biometrics
- Vol. 79 (3) , 1801-1813
Open AccessIntegrative Bayesian Models Using Post-Selective Inference: A Case Study in Radiogenomics
Integrative analyses based on statistically relevant associations between genomics and a wealth of intermediary phenotypes (such as imaging) provide vital insights into their clinical relevance in terms of the disease mechanisms. Estimates for uncertainty in the resulting integrative models are however unreliable unless inference accounts for the selection of these associations with accuracy. In this paper, we develop selection-aware Bayesian methods, which (1) counteract the impact of model selection bias through a “selection-aware posterior” in a flexible class of integrative Bayesian models post a selection of promising variables via ℓ1-regularized algorithms; (2) strike an inevitable trade-off between the quality of model selection and inferential power when the same data set is used for both selection and uncertainty estimation. Central to our methodological development, a carefully constructed conditional likelihood function deployed with a reparameterization mapping provides tractable updates when gradient-based Markov chain Monte Carlo (MCMC) sampling is used for estimating uncertainties from the selection-aware posterior. Applying our methods to a radiogenomic analysis, we successfully recover several important gene pathways and estimate uncertainties for their associations with patient survival times.
- Open Access
- 8 March 2022
- Published by Springer Nature in Scientific Reports
- Vol. 12 (1) , 1-11
GaWRDenMap: a quantitative framework to study the local variation in cell–cell interactions in pancreatic disease subtypes
Spatial pattern modelling concepts are being increasingly used in capturing disease heterogeneity. Quantification of heterogeneity in the tumor microenvironment is extremely important in pancreatic ductal adenocarcinoma (PDAC), which has been shown to co-occur with other pancreatic diseases and neoplasms with certain attributes that make visual discrimination difficult. In this paper, we propose the GaWRDenMap framework, that utilizes the concepts of geographically weighted regression (GWR) and a density function-based classification model, and apply it to a cohort of multiplex immunofluorescence images from patients belonging to six different pancreatic diseases. We used an internal cohort of 228 patients comprised of 34 Chronic Pancreatitis (CP), 71 PDAC, 70 intraductal papillary mucinous neoplasm (IPMN), 16 mucinous cystic neoplasm (MCN), 29 pancreatic intraductal neoplasia (PanIN) and 8 IPMN-associated PDAC patients. We utilized GWR to model the relationship between epithelial cells and immune cells on a spatial grid. The GWR model estimates were used to generate density signatures which were used in subsequent pairwise classification models to distinguish between any two pairs of disease groups. Image-level, as well as subject-level analysis, were performed. When applied to this dataset, our classification model showed significant discrimination ability in multiple pairwise comparisons, in comparison to commonly used abundance-based metrics, like the Morisita-Horn index. The model was able to best discriminate between CP and PDAC at both the subject- and image-levels. It was also able to reasonably discriminate between PDAC and IPMN. These results point to a potential difference in the spatial arrangement of epithelial and immune cells between CP, PDAC and IPMN, that could be of high diagnostic significance. Further validation on a more comprehensive dataset would be warranted.
- 8 March 2022
- Published by Springer Nature in Scientific Reports
- Vol. 12 (1) , 1-11
Open AccessGaWRDenMap: a quantitative framework to study the local variation in cell–cell interactions in pancreatic disease subtypes
Spatial pattern modelling concepts are being increasingly used in capturing disease heterogeneity. Quantification of heterogeneity in the tumor microenvironment is extremely important in pancreatic ductal adenocarcinoma (PDAC), which has been shown to co-occur with other pancreatic diseases and neoplasms with certain attributes that make visual discrimination difficult. In this paper, we propose the GaWRDenMap framework, that utilizes the concepts of geographically weighted regression (GWR) and a density function-based classification model, and apply it to a cohort of multiplex immunofluorescence images from patients belonging to six different pancreatic diseases. We used an internal cohort of 228 patients comprised of 34 Chronic Pancreatitis (CP), 71 PDAC, 70 intraductal papillary mucinous neoplasm (IPMN), 16 mucinous cystic neoplasm (MCN), 29 pancreatic intraductal neoplasia (PanIN) and 8 IPMN-associated PDAC patients. We utilized GWR to model the relationship between epithelial cells and immune cells on a spatial grid. The GWR model estimates were used to generate density signatures which were used in subsequent pairwise classification models to distinguish between any two pairs of disease groups. Image-level, as well as subject-level analysis, were performed. When applied to this dataset, our classification model showed significant discrimination ability in multiple pairwise comparisons, in comparison to commonly used abundance-based metrics, like the Morisita-Horn index. The model was able to best discriminate between CP and PDAC at both the subject- and image-levels. It was also able to reasonably discriminate between PDAC and IPMN. These results point to a potential difference in the spatial arrangement of epithelial and immune cells between CP, PDAC and IPMN, that could be of high diagnostic significance. Further validation on a more comprehensive dataset would be warranted.
- 11 November 2021
- Published by American Society of Neuroradiology (ASNR) in American Journal of Neuroradiology
- Vol. 43 (1) , 33-39
Quantifying T2-FLAIR Mismatch Using Geographically Weighted Regression and Predicting Molecular Status in Lower-Grade Gliomas
BACKGROUND AND PURPOSE: The T2-FLAIR mismatch sign is a validated imaging sign of isocitrate dehydrogenase–mutant 1p/19q noncodeleted gliomas. It is identified by radiologists through visual inspection of preoperative MR imaging scans and has been shown to identify isocitrate dehydrogenase–mutant 1p/19q noncodeleted gliomas with a high positive predictive value. We have developed an approach to quantify the T2-FLAIR mismatch signature and use it to predict the molecular status of lower-grade gliomas. MATERIALS AND METHODS: We used multiparametric MR imaging scans and segmentation labels of 108 preoperative lower-grade glioma tumors from The Cancer Imaging Archive. Clinical information and T2-FLAIR mismatch sign labels were obtained from supplementary material of relevant publications. We adopted an objective analytic approach to estimate this sign through a geographically weighted regression and used the residuals for each case to construct a probability density function (serving as a residual signature). These functions were then analyzed using an appropriate statistical framework. RESULTS: We observed statistically significant (P value = .05) differences between the averages of residual signatures for an isocitrate dehydrogenase–mutant 1p/19q noncodeleted class of tumors versus other categories. Our classifier predicts these cases with area under the curve of 0.98 and high specificity and sensitivity. It also predicts the T2-FLAIR mismatch sign within these cases with an under the curve of 0.93. CONCLUSIONS: On the basis of this retrospective study, we show that geographically weighted regression–based residual signatures are highly informative of the T2-FLAIR mismatch sign and can identify isocitrate dehydrogenase–mutation and 1p/19q codeletion status with high predictive power. The utility of the proposed quantification of the T2-FLAIR mismatch sign can be potentially validated through a prospective multi-institutional study.
- 11 November 2021
- Published by American Society of Neuroradiology (ASNR) in American Journal of Neuroradiology
- Vol. 43 (1) , 33-39
Quantifying T2-FLAIR Mismatch Using Geographically Weighted Regression and Predicting Molecular Status in Lower-Grade Gliomas
BACKGROUND AND PURPOSE: The T2-FLAIR mismatch sign is a validated imaging sign of isocitrate dehydrogenase–mutant 1p/19q noncodeleted gliomas. It is identified by radiologists through visual inspection of preoperative MR imaging scans and has been shown to identify isocitrate dehydrogenase–mutant 1p/19q noncodeleted gliomas with a high positive predictive value. We have developed an approach to quantify the T2-FLAIR mismatch signature and use it to predict the molecular status of lower-grade gliomas. MATERIALS AND METHODS: We used multiparametric MR imaging scans and segmentation labels of 108 preoperative lower-grade glioma tumors from The Cancer Imaging Archive. Clinical information and T2-FLAIR mismatch sign labels were obtained from supplementary material of relevant publications. We adopted an objective analytic approach to estimate this sign through a geographically weighted regression and used the residuals for each case to construct a probability density function (serving as a residual signature). These functions were then analyzed using an appropriate statistical framework. RESULTS: We observed statistically significant (P value = .05) differences between the averages of residual signatures for an isocitrate dehydrogenase–mutant 1p/19q noncodeleted class of tumors versus other categories. Our classifier predicts these cases with area under the curve of 0.98 and high specificity and sensitivity. It also predicts the T2-FLAIR mismatch sign within these cases with an under the curve of 0.93. CONCLUSIONS: On the basis of this retrospective study, we show that geographically weighted regression–based residual signatures are highly informative of the T2-FLAIR mismatch sign and can identify isocitrate dehydrogenase–mutation and 1p/19q codeletion status with high predictive power. The utility of the proposed quantification of the T2-FLAIR mismatch sign can be potentially validated through a prospective multi-institutional study.