Min Young Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Min Young Kim has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Benton Chuter
Department of Ophthalmology, Hamilton Eye Institute, University of Tennessee Health Science Center, Memphis, TN 38163, USA
4 shared publications - Monica M. Jablonski
Department of Ophthalmology, Hamilton Eye Institute, University of Tennessee Health Science Center, Memphis, TN 38163, USA
4 shared publications - T. J. Hollingsworth
Department of Ophthalmology, Hamilton Eye Institute, The University of Tennessee Health Science Center, Memphis, TN 38163, USA
4 shared publications - Abraham A. Palmer
Department of Psychiatry, University of California San Diego
2 shared publications - Gideon Wall
College of Medicine, University of Tennessee Health Science Center, Memphis, TN 38163, USA
2 shared publications - Hao Chen
Department of Pharmacology, Addiction Science and Toxicology, University of Tennessee Health Science Center, Memphis, TN
2 shared publications
Latest Publications
- Open Access
A Systematic Review and Independent Benchmarking of Automated Nerve Morphometry Methods
Objective: To systematically review automated nerve morphometry tools and independently benchmark their performance on independent optic nerve datasets. Design: Systematic review and comparative benchmarking study. Controls: Benchmarking was performed using paraphenylenediamine-stained mouse (n = 85) and rat (n = 44) optic nerve images with manually annotated axon counts as ground truth. Methods: Published studies describing automated or semi-automated neural tissue morphometry tools were identified through systematic searches of PubMed, Embase, and Scopus through January 2026 following PRISMA guidelines. Data extraction covered 70 fields across tool capabilities, imaging modality, species, automation level, and validation approach. Eighteen eligible tools (8 deep learning [DL], 10 classical computer vision [CV]) were benchmarked on both mouse and rat independent datasets. Main Outcome Measures: Performance was assessed by mean absolute percentage error (MAPE), Pearson correlation, and median predicted-to-ground-truth ratio. Tools were ranked per image and compared using Friedman tests with Nemenyi post-hoc analysis. Results: Seventy-one studies met inclusion criteria, spanning from 1999 to 2026. Deep learning methods represented 38% (27/71) of studies, increasing from 0% before 2017 to over 55% of publications after 2020. Axon counting was the most common output (73%, 52/71), while only 35% (25/71) reported g-ratio. Among benchmarked tools, Marina (CV, 2010) achieved the lowest average MAPE (32.9%). The top five tools (MAPE ranging from 32.9 to 44.8%) included both CV and DL methods and were statistically indistinguishable by Friedman-Nemenyi analysis (p > 0.05). Performance varied substantially across datasets: AxonJ (CV) achieved the second best MAPE on rat images (27.7%) but the worst on mouse images (438.6%). Conclusions: No single tool demonstrated consistently superior performance across both datasets. Classical and deep learning approaches achieved comparable accuracy for axon counting. Tool selection should be guided by target species, tissue preparation protocol, and desired morphometric outputs. This systematic review and independent benchmarking study provide an evidence base for tool selection in optic nerve research.
Open AccessA Systematic Review and Independent Benchmarking of Automated Nerve Morphometry Methods
Objective: To systematically review automated nerve morphometry tools and independently benchmark their performance on independent optic nerve datasets. Design: Systematic review and comparative benchmarking study. Controls: Benchmarking was performed using paraphenylenediamine-stained mouse (n = 85) and rat (n = 44) optic nerve images with manually annotated axon counts as ground truth. Methods: Published studies describing automated or semi-automated neural tissue morphometry tools were identified through systematic searches of PubMed, Embase, and Scopus through January 2026 following PRISMA guidelines. Data extraction covered 70 fields across tool capabilities, imaging modality, species, automation level, and validation approach. Eighteen eligible tools (8 deep learning [DL], 10 classical computer vision [CV]) were benchmarked on both mouse and rat independent datasets. Main Outcome Measures: Performance was assessed by mean absolute percentage error (MAPE), Pearson correlation, and median predicted-to-ground-truth ratio. Tools were ranked per image and compared using Friedman tests with Nemenyi post-hoc analysis. Results: Seventy-one studies met inclusion criteria, spanning from 1999 to 2026. Deep learning methods represented 38% (27/71) of studies, increasing from 0% before 2017 to over 55% of publications after 2020. Axon counting was the most common output (73%, 52/71), while only 35% (25/71) reported g-ratio. Among benchmarked tools, Marina (CV, 2010) achieved the lowest average MAPE (32.9%). The top five tools (MAPE ranging from 32.9 to 44.8%) included both CV and DL methods and were statistically indistinguishable by Friedman-Nemenyi analysis (p > 0.05). Performance varied substantially across datasets: AxonJ (CV) achieved the second best MAPE on rat images (27.7%) but the worst on mouse images (438.6%). Conclusions: No single tool demonstrated consistently superior performance across both datasets. Classical and deep learning approaches achieved comparable accuracy for axon counting. Tool selection should be guided by target species, tissue preparation protocol, and desired morphometric outputs. This systematic review and independent benchmarking study provide an evidence base for tool selection in optic nerve research.
- Open Access
- 30 May 2026
- Published by MDPI AG in Bioengineering
- Vol. 13 (6) , 647
Comparison of Deep Learning Tools for Optic Nerve Axon Quantification Finds Limited Generalizability Upon Independent Validation
Machine learning approaches for automated quantification of optic nerve histology have emerged as tools for the objective assessment of axonal injury in experimental glaucoma models. However, their generalizability to independent datasets remains unclear. Guided by a scoping review following PRISMA-ScR guidelines, this study evaluated the performance of publicly available models on novel datasets. PubMed, EMBASE, Scopus, and Cochrane CENTRAL were searched (2000–2025). Two reviewers screened studies and extracted model characteristics and performance metrics. Three models (AxoNet, AxonDeepSeg, and AxoNet 2.0) were independently validated on a rat optic nerve dataset (44 images; 6941 axons) and a mouse optic nerve dataset (74 full cross-sections). From 2036 records, four manuscripts describing three deep learning models met the inclusion criteria, with reported correlation coefficients of 0.959–0.99 between model predictions and reference counts. On the rat dataset, the performance correlation declined (r = 0.831–0.907), precision remained high (>0.94), but recall was low (0.18–0.27), with Dice coefficients of 0.29–0.40. On the mouse dataset, correlations decreased further (r = 0.57–0.74) and model rankings differed, reflecting the domain shift and scale-dependent effects. These findings demonstrate strong within-study performance but reduced generalizability to independent datasets, highlighting the need for standardized validation datasets and multi-center testing.
- 30 May 2026
- Published by MDPI AG in Bioengineering
- Vol. 13 (6) , 647
Open AccessComparison of Deep Learning Tools for Optic Nerve Axon Quantification Finds Limited Generalizability Upon Independent Validation
Machine learning approaches for automated quantification of optic nerve histology have emerged as tools for the objective assessment of axonal injury in experimental glaucoma models. However, their generalizability to independent datasets remains unclear. Guided by a scoping review following PRISMA-ScR guidelines, this study evaluated the performance of publicly available models on novel datasets. PubMed, EMBASE, Scopus, and Cochrane CENTRAL were searched (2000–2025). Two reviewers screened studies and extracted model characteristics and performance metrics. Three models (AxoNet, AxonDeepSeg, and AxoNet 2.0) were independently validated on a rat optic nerve dataset (44 images; 6941 axons) and a mouse optic nerve dataset (74 full cross-sections). From 2036 records, four manuscripts describing three deep learning models met the inclusion criteria, with reported correlation coefficients of 0.959–0.99 between model predictions and reference counts. On the rat dataset, the performance correlation declined (r = 0.831–0.907), precision remained high (>0.94), but recall was low (0.18–0.27), with Dice coefficients of 0.29–0.40. On the mouse dataset, correlations decreased further (r = 0.57–0.74) and model rankings differed, reflecting the domain shift and scale-dependent effects. These findings demonstrate strong within-study performance but reduced generalizability to independent datasets, highlighting the need for standardized validation datasets and multi-center testing.
- Open Access
Visual Function Correlates More Strongly with Glial Coverage than Axon Count Across Multiple Mouse Strains
Objective: To determine whether non-axon optic nerve morphometric features correlate with clinical visual function as strongly as the traditional axon count gold standard. Design: Cross-sectional histological analysis with longitudinal clinical correlation. Subjects: Eighteen mice from three strains: C57BL/6J (n=6), BXD51 (n=6), and DBA/2J (n=6). Methods: Left eye (OS) optic nerves from mice euthanized at 12 months of age were resin-embedded and stained with p-phenylenediamine. Bright-field cross-sectional images were segmented using an AxonDeepSeg-based workflow to generate axon, myelin, whole nerve, and glial coverage masks for morphometric quantification. Seven morphometrics were extracted: axon count (nAx), axon density (AxDen), glial coverage area ratio (GliaR), mean solidity (Sol), mean axon diameter (AxDiam), mean myelin area (MyArea), and mean axon-myelin area (AxMyArea). Morphometrics were correlated with longitudinal clinical data collected at 1, 3, 6, 9, and 12 months, including visual acuity (VA), contrast threshold, intraocular pressure (IOP), and pattern electroretinography P50 and N95 amplitudes (PERG P50 and N95). Main Outcome Measures: Pearson correlation coefficients were used to assess associations between morphometric features and clinical measures, and Fisher z-transformed meta-analytic correlations were used to aggregate these associations across ages. Results: VA and contrast threshold demonstrated strong correlations with GliaR that matched or exceeded nAx. Meta-analysis across ages revealed GliaR correlated with VA (r = -0.84, p = 4.49 × 10 -21 ) and contrast threshold (r = 0.86, p =7.55 × 10- 23 ), comparable to nAx correlations with VA (r = 0.80, p =8.13×10 -17 ) and contrast threshold (r = -0.80, p= 1.74×10 -16 ). Structure-function relationships shifted with age: at 6 months, GliaR had the strongest correlation with contrast threshold (r = 0.96), while at 12 months, AxDiam became the dominant correlate of both VA (r = 0.77) and contrast threshhold (r = -0.74). IOP, PERG P50, and PERG N95 exhibited weak correlations with all morphometrics (|r| < 0.27). Conclusions: Non-axon morphometrics, particularly glial coverage area ratio, correlate with visual function as strongly as traditional axon count. Automated optic nerve assessment should incorporate glial and other non-axon features. Further, stage-aware biomarker selection may better capture structure-function relationships in glaucoma.
Open AccessVisual Function Correlates More Strongly with Glial Coverage than Axon Count Across Multiple Mouse Strains
Objective: To determine whether non-axon optic nerve morphometric features correlate with clinical visual function as strongly as the traditional axon count gold standard. Design: Cross-sectional histological analysis with longitudinal clinical correlation. Subjects: Eighteen mice from three strains: C57BL/6J (n=6), BXD51 (n=6), and DBA/2J (n=6). Methods: Left eye (OS) optic nerves from mice euthanized at 12 months of age were resin-embedded and stained with p-phenylenediamine. Bright-field cross-sectional images were segmented using an AxonDeepSeg-based workflow to generate axon, myelin, whole nerve, and glial coverage masks for morphometric quantification. Seven morphometrics were extracted: axon count (nAx), axon density (AxDen), glial coverage area ratio (GliaR), mean solidity (Sol), mean axon diameter (AxDiam), mean myelin area (MyArea), and mean axon-myelin area (AxMyArea). Morphometrics were correlated with longitudinal clinical data collected at 1, 3, 6, 9, and 12 months, including visual acuity (VA), contrast threshold, intraocular pressure (IOP), and pattern electroretinography P50 and N95 amplitudes (PERG P50 and N95). Main Outcome Measures: Pearson correlation coefficients were used to assess associations between morphometric features and clinical measures, and Fisher z-transformed meta-analytic correlations were used to aggregate these associations across ages. Results: VA and contrast threshold demonstrated strong correlations with GliaR that matched or exceeded nAx. Meta-analysis across ages revealed GliaR correlated with VA (r = -0.84, p = 4.49 × 10 -21 ) and contrast threshold (r = 0.86, p =7.55 × 10- 23 ), comparable to nAx correlations with VA (r = 0.80, p =8.13×10 -17 ) and contrast threshold (r = -0.80, p= 1.74×10 -16 ). Structure-function relationships shifted with age: at 6 months, GliaR had the strongest correlation with contrast threshold (r = 0.96), while at 12 months, AxDiam became the dominant correlate of both VA (r = 0.77) and contrast threshhold (r = -0.74). IOP, PERG P50, and PERG N95 exhibited weak correlations with all morphometrics (|r| < 0.27). Conclusions: Non-axon morphometrics, particularly glial coverage area ratio, correlate with visual function as strongly as traditional axon count. Automated optic nerve assessment should incorporate glial and other non-axon features. Further, stage-aware biomarker selection may better capture structure-function relationships in glaucoma.
- Open Access
Comparison of Deep Learning Tools for Optic Nerve Axon Quantification Finds Limited Generalizability on Independent Validation
Purpose: Machine learning approaches for automated quantification of optic nerve histology have emerged as potential tools for objective assessment of axonal injury in experimental glaucoma models. However, the generalizability of these models to independent datasets remains unclear. Guided by a scoping review of the literature, this study performed independent validation testing of publicly available models on a novel rat optic nerve dataset to assess their generalizability. Methods: We conducted a scoping review following PRISMA-ScR guidelines. PubMed, EMBASE, Scopus, and Cochrane CENTRAL were searched from 2000 through 2025. Two reviewers independently screened records and extracted data on model characteristics and performance metrics. Additionally, we performed independent validation of three models (AxoNet, AxonDeepSeg, AxoNet 2.0) on a novel rat optic nerve dataset comprising 57 images with 9,514 manually annotated axons. Because AxonDeep is not publicly available, we instead evaluated AxonDeepSeg, a separate publicly available deep learning-based tool that, while not previously applied to optic nerve tissue, is widely used for nerve fiber segmentation. Results: From 2,036 records, four manuscripts describing three deep learning models met inclusion criteria. Published correlation coefficients between model predictions and reference counts ranged from 0.959 to 0.99. On independent validation, performance was reduced: AxoNet 2.0 achieved the highest correlation (r = 0.89), followed by AxonDeepSeg (r = 0.86) and AxoNet (r = 0.79). Segmentation quality metrics revealed high precision (>0.94) but low recall (0.18 to 0.27), with Dice coefficients of 0.29 to 0.40, substantially below published benchmarks of 0.81. Conclusions: Deep learning models for optic nerve histology demonstrate strong within-study performance but show meaningful performance decrements when applied to independent datasets. The observed generalizability gap (correlations 0.07 to 0.182 points below published values) demonstrates the need for standardized validation datasets and multi-center testing before widespread adoption of these tools.
Open AccessComparison of Deep Learning Tools for Optic Nerve Axon Quantification Finds Limited Generalizability on Independent Validation
Purpose: Machine learning approaches for automated quantification of optic nerve histology have emerged as potential tools for objective assessment of axonal injury in experimental glaucoma models. However, the generalizability of these models to independent datasets remains unclear. Guided by a scoping review of the literature, this study performed independent validation testing of publicly available models on a novel rat optic nerve dataset to assess their generalizability. Methods: We conducted a scoping review following PRISMA-ScR guidelines. PubMed, EMBASE, Scopus, and Cochrane CENTRAL were searched from 2000 through 2025. Two reviewers independently screened records and extracted data on model characteristics and performance metrics. Additionally, we performed independent validation of three models (AxoNet, AxonDeepSeg, AxoNet 2.0) on a novel rat optic nerve dataset comprising 57 images with 9,514 manually annotated axons. Because AxonDeep is not publicly available, we instead evaluated AxonDeepSeg, a separate publicly available deep learning-based tool that, while not previously applied to optic nerve tissue, is widely used for nerve fiber segmentation. Results: From 2,036 records, four manuscripts describing three deep learning models met inclusion criteria. Published correlation coefficients between model predictions and reference counts ranged from 0.959 to 0.99. On independent validation, performance was reduced: AxoNet 2.0 achieved the highest correlation (r = 0.89), followed by AxonDeepSeg (r = 0.86) and AxoNet (r = 0.79). Segmentation quality metrics revealed high precision (>0.94) but low recall (0.18 to 0.27), with Dice coefficients of 0.29 to 0.40, substantially below published benchmarks of 0.81. Conclusions: Deep learning models for optic nerve histology demonstrate strong within-study performance but show meaningful performance decrements when applied to independent datasets. The observed generalizability gap (correlations 0.07 to 0.182 points below published values) demonstrates the need for standardized validation datasets and multi-center testing before widespread adoption of these tools.
- Open Access
- 14 October 2025
- Published by MDPI AG in Biomolecules
- Vol. 15 (10) , 1450
Alterations of Bioactive Lipid Profiles in the Retina Following Traumatic Optic Neuropathy in Mice
Traumatic optic neuropathy (TON) causes vision loss through compression and contusion, yet there is no consensus on the most effective treatment. Polyunsaturated fatty acid (PUFA)-derived bioactive lipids metabolized by lipoxygenase (LOX), cytochrome P450 (CYP), and cyclooxygenase (COX) enzymes are known mediators of inflammation and neurodegeneration. However, their role in TON-related retinal pathology remains unclear. Controlled orbital impact (COI) was used to induce unilateral TON in mice with controlled velocity (2–3 m/s), with the fellow eye serving as an internal control. Retina tissues were collected three days post-injury and analyzed by LC/MS to quantify bioactive lipid metabolites from ω−6 and ω−3 PUFAs. Statistical analysis was performed using paired, nonparametric Wilcoxon signed-rank tests with Benjamini–Hochberg false discovery rate (FDR) correction. Results showed that among 38 reliably detected metabolites, no individual lipid showed a statistically significant difference between TON and control eyes after FDR correction (q < 0.05). However, both individual and pathway-level analysis revealed consistent trends toward increased expression of LOX- and CYP-derived metabolites across FDA PUFA substrates, including arachidonic acid (AA), linoleic acid (LA), and docosahexaenoic acid (DHA). These findings support further investigation into lipid-mediated inflammation in TON and its potential as a therapeutic target, particularly through expanding both the sample size and the post-TON time periods.
- 14 October 2025
- Published by MDPI AG in Biomolecules
- Vol. 15 (10) , 1450
Open AccessAlterations of Bioactive Lipid Profiles in the Retina Following Traumatic Optic Neuropathy in Mice
Traumatic optic neuropathy (TON) causes vision loss through compression and contusion, yet there is no consensus on the most effective treatment. Polyunsaturated fatty acid (PUFA)-derived bioactive lipids metabolized by lipoxygenase (LOX), cytochrome P450 (CYP), and cyclooxygenase (COX) enzymes are known mediators of inflammation and neurodegeneration. However, their role in TON-related retinal pathology remains unclear. Controlled orbital impact (COI) was used to induce unilateral TON in mice with controlled velocity (2–3 m/s), with the fellow eye serving as an internal control. Retina tissues were collected three days post-injury and analyzed by LC/MS to quantify bioactive lipid metabolites from ω−6 and ω−3 PUFAs. Statistical analysis was performed using paired, nonparametric Wilcoxon signed-rank tests with Benjamini–Hochberg false discovery rate (FDR) correction. Results showed that among 38 reliably detected metabolites, no individual lipid showed a statistically significant difference between TON and control eyes after FDR correction (q < 0.05). However, both individual and pathway-level analysis revealed consistent trends toward increased expression of LOX- and CYP-derived metabolites across FDA PUFA substrates, including arachidonic acid (AA), linoleic acid (LA), and docosahexaenoic acid (DHA). These findings support further investigation into lipid-mediated inflammation in TON and its potential as a therapeutic target, particularly through expanding both the sample size and the post-TON time periods.
- Open Access
- 16 September 2025
- Published by Wiley in Cancer Medicine
- Vol. 14 (18) , e71253
“Quite Devastating, Traumatic, and Lonely”: Understanding the Mental Health Experiences of Hispanic/Latino Rectal Cancer Patients and Their Caregivers
Background: Psychological distress increases after a cancer diagnosis. However, patients who identify as Hispanic/Latino are less likely than their White counterparts to receive mental health care after being diagnosed, placing them at risk of poor psychological adjustment. The parent research objective aimed to characterize the treatment experiences of rectal cancer patients who identify as Hispanic/Latino. This qualitative secondary analysis specifically explored patients' emotional and psychological experiences. Methods: We partnered with local community health workers to recruit rectal cancer patients who identified as Hispanic/Latino and their caregivers from California's Bay Area and Central Valley. Semi‐structured interviews ( n = 21) were conducted in either Spanish or English based on patient preference. Initial themes were identified using reflexive analysis. Experiences of distress were found in nearly all interviews, motivating this qualitative secondary analysis using directed content analysis. Results: We identified three main themes: (1) emotional reactions to the initial diagnosis, (2) perspectives and emotions experienced throughout treatment, and (3) coping strategies and advice for others. Participants reported shock and incongruence with their diagnosis. The treatment journey caused feelings of embarrassment and frustration from difficulties such as racial profiling and insurance. Patients coped through faith and support from trusted individuals. Conclusions: Rectal cancer patients who identify as Hispanic/Latino endure emotional distress at multiple stages of their cancer journey, including from sources that are unique to this population. Culturally and linguistically tailored interventions are needed to help mitigate this psychological burden and to provide active, coordinated support throughout the treatment phase in order to better prepare patients for survivorship.
- 16 September 2025
- Published by Wiley in Cancer Medicine
- Vol. 14 (18) , e71253
Open Access“Quite Devastating, Traumatic, and Lonely”: Understanding the Mental Health Experiences of Hispanic/Latino Rectal Cancer Patients and Their Caregivers
Background: Psychological distress increases after a cancer diagnosis. However, patients who identify as Hispanic/Latino are less likely than their White counterparts to receive mental health care after being diagnosed, placing them at risk of poor psychological adjustment. The parent research objective aimed to characterize the treatment experiences of rectal cancer patients who identify as Hispanic/Latino. This qualitative secondary analysis specifically explored patients' emotional and psychological experiences. Methods: We partnered with local community health workers to recruit rectal cancer patients who identified as Hispanic/Latino and their caregivers from California's Bay Area and Central Valley. Semi‐structured interviews ( n = 21) were conducted in either Spanish or English based on patient preference. Initial themes were identified using reflexive analysis. Experiences of distress were found in nearly all interviews, motivating this qualitative secondary analysis using directed content analysis. Results: We identified three main themes: (1) emotional reactions to the initial diagnosis, (2) perspectives and emotions experienced throughout treatment, and (3) coping strategies and advice for others. Participants reported shock and incongruence with their diagnosis. The treatment journey caused feelings of embarrassment and frustration from difficulties such as racial profiling and insurance. Patients coped through faith and support from trusted individuals. Conclusions: Rectal cancer patients who identify as Hispanic/Latino endure emotional distress at multiple stages of their cancer journey, including from sources that are unique to this population. Culturally and linguistically tailored interventions are needed to help mitigate this psychological burden and to provide active, coordinated support throughout the treatment phase in order to better prepare patients for survivorship.
- Open Access
- 22 July 2024
- Published by Elsevier in American Journal of Ophthalmology Case Reports
- Vol. 36, 102124
Siponimod-associated cystoid macular edema without known risk factors
- 22 July 2024
- Published by Elsevier in American Journal of Ophthalmology Case Reports
- Vol. 36, 102124
Open AccessSiponimod-associated cystoid macular edema without known risk factors