1
|
Chung MK, Ramos CG, De Paiva FB, Mathis J, Prabhakaran V, Nair VA, Meyerand ME, Hermann BP, Binder JR, Struck AF. Unified topological inference for brain networks in temporal lobe epilepsy using the Wasserstein distance. Neuroimage 2023; 284:120436. [PMID: 37931870 PMCID: PMC11074922 DOI: 10.1016/j.neuroimage.2023.120436] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/15/2023] [Revised: 09/14/2023] [Accepted: 10/30/2023] [Indexed: 11/08/2023] Open
Abstract
Persistent homology offers a powerful tool for extracting hidden topological signals from brain networks. It captures the evolution of topological structures across multiple scales, known as filtrations, thereby revealing topological features that persist over these scales. These features are summarized in persistence diagrams, and their dissimilarity is quantified using the Wasserstein distance. However, the Wasserstein distance does not follow a known distribution, posing challenges for the application of existing parametric statistical models. To tackle this issue, we introduce a unified topological inference framework centered on the Wasserstein distance. Our approach has no explicit model and distributional assumptions. The inference is performed in a completely data driven fashion. We apply this method to resting-state functional magnetic resonance images (rs-fMRI) of temporal lobe epilepsy patients collected from two different sites: the University of Wisconsin-Madison and the Medical College of Wisconsin. Importantly, our topological method is robust to variations due to sex and image acquisition, obviating the need to account for these variables as nuisance covariates. We successfully localize the brain regions that contribute the most to topological differences. A MATLAB package used for all analyses in this study is available at https://github.com/laplcebeltrami/PH-STAT.
Collapse
Affiliation(s)
- Moo K Chung
- Department of Biostatistics and Medical Informatics, University of Wisconsin-Madison, USA.
| | | | | | | | | | - Veena A Nair
- Department of Radiology, University of Wisconsin-Madison, USA.
| | - Mary E Meyerand
- Departments of Medical Physics & Biomedical Engineering, University of Wisconsin-Madison, USA.
| | - Bruce P Hermann
- Department of Neurology, University of Wisconsin-Madison, USA.
| | | | - Aaron F Struck
- Department of Neurology, University of Wisconsin-Madison, USA.
| |
Collapse
|
2
|
Chung MK, Ramos CG, De Paiva FB, Mathis J, Prabharakaren V, Nair VA, Meyerand E, Hermann BP, Binder JR, Struck AF. Unified Topological Inference for Brain Networks in Temporal Lobe Epilepsy Using the Wasserstein Distance. ArXiv 2023:arXiv:2302.06673v3. [PMID: 36824424 PMCID: PMC9949148] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Grants] [Download PDF] [Subscribe] [Scholar Register] [Indexed: 02/25/2023]
Abstract
Persistent homology offers a powerful tool for extracting hidden topological signals from brain networks. It captures the evolution of topological structures across multiple scales, known as filtrations, thereby revealing topological features that persist over these scales. These features are summarized in persistence diagrams, and their dissimilarity is quantified using the Wasserstein distance. However, the Wasserstein distance does not follow a known distribution, posing challenges for the application of existing parametric statistical models. To tackle this issue, we introduce a unified topological inference framework centered on the Wasserstein distance. Our approach has no explicit model and distributional assumptions. The inference is performed in a completely data driven fashion. We apply this method to resting-state functional magnetic resonance images (rs-fMRI) of temporal lobe epilepsy patients collected from two different sites: the University of Wisconsin-Madison and the Medical College of Wisconsin. Importantly, our topological method is robust to variations due to sex and image acquisition, obviating the need to account for these variables as nuisance covariates. We successfully localize the brain regions that contribute the most to topological differences. A MATLAB package used for all analyses in this study is available at https://github.com/laplcebeltrami/PH-STAT.
Collapse
Affiliation(s)
- Moo K Chung
- Department of Biostatistics and Medical Informatics, University of Wisconsin-Madison, USA
| | | | | | | | | | - Veena A Nair
- Department of Radiology, University of Wisconsin-Madison, USA
| | - Elizabeth Meyerand
- Departments of Medical Physics & Biomedical Engineering, University of Wisconsin-Madison, USA
| | - Bruce P Hermann
- Department of Neurology, University of Wisconsin-Madison, USA
| | | | - Aaron F Struck
- Department of Neurology, University of Wisconsin-Madison, USA
| |
Collapse
|
3
|
Zeng H, Ramos CG, Nair VA, Hu Y, Liao J, La C, Chen L, Gan Y, Wen F, Hermann B, Prabhakaran V. Regional homogeneity (ReHo) changes in new onset versus chronic benign epilepsy of childhood with centrotemporal spikes (BECTS): A resting state fMRI study. Epilepsy Res 2015; 116:79-85. [PMID: 26354170 PMCID: PMC4567694 DOI: 10.1016/j.eplepsyres.2015.06.017] [Citation(s) in RCA: 34] [Impact Index Per Article: 3.8] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 01/28/2015] [Revised: 06/11/2015] [Accepted: 06/23/2015] [Indexed: 02/05/2023]
Abstract
OBJECTIVE The purpose of this study was to investigate regional homogeneity (ReHo) in children with new-onset drug-naïve Benign Epilepsy with Centrotemporal Spikes (BECTS), chronic BECTS and healthy controls (HC) using the Regional Homogeneity (ReHo) method applied to resting state fMRI data. METHODS Resting state fMRI data was collected from three groups of children aged 6-13, including new onset drug naïve BECTS, chronic BECTS with medication, and HC; the data analyzed by ReHo method. Mandarin school exams scores were acquired and compared across groups. RESULTS There were three main findings. Firstly, compared with HC, abnormally increased ReHo was observed in bilateral sensorimotor regions in new onset BECTS which normalized or even reversed in the chronic BECTS group. Secondly, enhanced ReHo was found in the left frontal language region in the two BECTS groups, with even higher ReHo value in the chronic group. Lastly, decreased ReHo was found in regions of the default mode network (DMN), bilateral occipital lobes and cerebellum in both the new onset and chronic BECTS groups, lower in chronic BECTS. Behavioral analyses of school scores showed the chronic BECTS group presented significantly lower scores compared to HC (p<.05). SIGNIFICANCE The coherence of low frequency fluctuations is disrupted in sensorimotor, language and DMN-related regions in new-onset BECTS. Some of these effects seem to be selectively normalized in chronic BECTS, thus allowing us to explore possible chronicity and AED-induced effects on BECTS. Abnormal ReHo in left language and DMN regions could be responsible for impairments of cognitive function.
Collapse
Affiliation(s)
- Hongwu Zeng
- Department of Radiology, Shenzhen Children's Hospital, Guangdong, China; Department of Radiology, University of Wisconsin, Madison, USA.
| | | | - Veena A Nair
- Department of Radiology, University of Wisconsin, Madison, USA.
| | - Yan Hu
- Department of Neurology, Shenzhen Children's Hospital, Guangdong, China.
| | - Jianxiang Liao
- Department of Neurology, Shenzhen Children's Hospital, Guangdong, China.
| | - Christian La
- Department of Radiology, University of Wisconsin, Madison, USA; Neurosciences Training Program, University of Wisconsin, Madison, USA.
| | - Li Chen
- Department of Neurology, Shenzhen Children's Hospital, Guangdong, China.
| | - Yungen Gan
- Department of Radiology, Shenzhen Children's Hospital, Guangdong, China.
| | - Feiqiu Wen
- Department of Neurology, Shenzhen Children's Hospital, Guangdong, China.
| | - Bruce Hermann
- Department of Neurology, University of Wisconsin, Madison, USA.
| | - Vivek Prabhakaran
- Department of Radiology, University of Wisconsin, Madison, USA; Department of Medical Physics, University of Wisconsin, Madison, USA; Neurosciences Training Program, University of Wisconsin, Madison, USA; Department of Neurology, University of Wisconsin, Madison, USA.
| |
Collapse
|
4
|
Carvalhal GF, Humphrey PA, Thorson P, Yan Y, Ramos CG, Catalona WJ. Visual estimate of the percentage of carcinoma is an independent predictor of prostate carcinoma recurrence after radical prostatectomy. Cancer 2000; 89:1308-14. [PMID: 11002227 DOI: 10.1002/1097-0142(20000915)89:6<1308::aid-cncr16>3.0.co;2-3] [Citation(s) in RCA: 68] [Impact Index Per Article: 2.8] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/09/2022]
Abstract
BACKGROUND The importance of tumor size measurements as predictors of disease recurrence after radical prostatectomy in patients with prostate carcinoma remains somewhat controversial, and many pathologists do not report these measurements routinely. The authors studied the correlation between the visual estimate of the percentage of carcinoma in prostatic tissue from radical prostatectomy specimens and prostate carcinoma recurrence rates in a series of 595 patients who underwent radical prostatectomy. METHODS A total of 595 men with clinically localized prostate carcinoma were treated by the same surgeon (W.J.C.) from 1993 through 1997. The percentage of carcinoma in radical prostatectomy specimens from these patients was assessed microscopically through visual estimate. The authors used Kaplan-Meier product limit estimates, log-rank statistics, and the Cox proportional hazards model to evaluate the percentage of carcinoma in the pathologic specimens as predictors of recurrence free survival. RESULTS Of the 595 patients, 46 (8%) had evidence of tumor recurrence. The mean percentage of carcinoma in the prostatectomy specimen was 11.3% in the group of patients who did not have disease recurrence and 23.8% in the group of patients who did experience disease recurrence. The percentage of carcinoma, preoperative prostate specific antigen levels, tumor differentiation (histologic Gleason grade), and pathologic stage all were significant predictors of disease recurrence according to the Kaplan-Meier method (all log-rank P values < 0.0001). Using the Cox proportional hazards model, controlling for all of these variables, only pathologic tumor stage, Gleason score, and percentage of carcinoma proved to be independent predictors of disease recurrence. In the final model, which included pathologic stage, Gleason score, and percentage of carcinoma, for each 5% increase in the percentage of carcinoma in the surgical specimen, there was an 11% incremental increase in the chance of tumor recurrence. CONCLUSIONS The visual estimate of the percentage of carcinoma in prostatic tissue specimens from patients who undergo radical prostatectomy is a practical, simple, and inexpensive method that provides important prognostic information after radical prostatectomy.
Collapse
Affiliation(s)
- G F Carvalhal
- Division of Urologic Surgery, Washington University School of Medicine, St. Louis, Missouri 63110, USA
| | | | | | | | | | | |
Collapse
|
5
|
|
6
|
Ramos CG, Carvahal GF, Mager DE, Haberer B, Catalona WJ. The effect of high grade prostatic intraepithelial neoplasia on serum total and percentage of free prostate specific antigen levels. J Urol 1999; 162:1587-90. [PMID: 10524873] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [MESH Headings] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/14/2023]
Abstract
PURPOSE It is established that the percentage of free prostate specific antigen (PSA) in serum is low in patients with prostate cancer. An unanswered question is whether a low percentage of free PSA can be explained by high grade prostatic intraepithelial neoplasia alone. We compared the percentage of free PSA in men with high grade prostatic intraepithelial neoplasia alone, prostate cancer, benign prostatic hyperplasia (BPH) and a normal prostate (that is normal digital rectal examination and PSA less than or equal to 2.5 ng./ml.). MATERIALS AND METHODS From October 1994 through December 1997, 48 men were diagnosed with high grade prostatic intraepithelial neoplasia without concomitant prostate cancer. Of these men 43 with a mean age plus or minus standard deviation of 67.4 +/- 7.8 years comprised our study group. To date none has been diagnosed with cancer during followup. Serum free and total PSA levels were measured, and the percentage of free PSA was calculated. The percentage of free PSA in the 43 men was compared to that in 50 with prostate cancer (mean age 65.4 +/- 7.8 years), 50 with biopsy proved BPH (67 +/- 7) and 43 with a normal prostate (61 +/- 8). RESULTS There was no significant difference in mean total serum PSA in patients with high grade prostatic intraepithelial neoplasia, prostate cancer or BPH. The percentage of free PSA was significantly lower in patients with prostate cancer (14.9 +/- 6.5%) than those with high grade prostatic intraepithelial neoplasia (20.8 +/- 7.1%), BPH (20.1 +/- 7.3%) or a normal prostate (27.7 +/- 12.2%). There was also no significant difference in the percentage of free PSA between men with high grade prostatic intraepithelial neoplasia (20.8 +/- 7.1%) and those with BPH (20.1 +/- 7.3%). Additionally, men with a normal prostate had a higher percentage of free PSA (27.7%) than those with BPH (20.1%), high grade prostatic intraepithelial neoplasia (20.8%) or prostate cancer (14.9%). CONCLUSIONS The percentages of free PSA in men with high grade prostatic intraepithelial neoplasia and BPH are similar, and significantly higher than those found in men with prostate cancer.
Collapse
Affiliation(s)
- C G Ramos
- Department of Surgery, Washington University School of Medicine, St. Louis, Missouri, USA
| | | | | | | | | |
Collapse
|
7
|
Abstract
With current clinical practice, most newly diagnosed cases of prostate cancer are potentially life-threatening yet still curable. The anatomical (nerve-sparing) radical prostatectomy has dramatically improved the results of surgical treatment. Other new management options, including conformal (three-dimensional) external beam radiation therapy, radioactive seed implantation (brachytherapy), cryoablation, and hormonal therapy, may be useful in some patients, but they are all probably less effective than radical prostatectomy. Suitability for radical prostatectomy generally requires a clinically localized, potentially life-threatening tumor [as defined by Gleason grade, tumor stage, and serum prostate-specific antigen (PSA) level], a life expectancy of 10 years, and no serious co-morbid medical conditions. With contemporary radical prostatectomy, about 70% of men with clinically localized disease will be cured, depending on tumor grade, tumor stage, and the serum PSA level. Urinary continence and sexual potency can be preserved in most patients, but substantially better results have been reported from centers of excellence than from community-based series. Other complications occur in about 10% of patients and with greater frequency in older patients. The operative mortality rate is less than 0.5%. Neoadjuvant hormonal therapy does not appear to affect treatment failure rates in patients undergoing radical prostatectomy. Prostatectomy may be beneficial in patients with microscopic lymph node metastases. Postoperative adjuvant radiotherapy may also be beneficial for patients with adverse pathologic findings. Salvage radical prostatectomy after radiation failure is associated with a 10-fold higher risk of complications and limited prospects for cure. Prospective, randomized clinical trials are underway to compare the results of radical prostatectomy with other treatments. Currently, radical prostatectomy is considered the preferred treatment for men with localized disease and a 10-year life expectancy.
Collapse
Affiliation(s)
- W J Catalona
- Division of Urologic Surgery, Department of Surgery, Washington University School of Medicine, St. Louis, MO, USA
| | | | | |
Collapse
|
8
|
Ramos CG, Carvalhal GF, Smith DS, Mager DE, Catalona WJ. Clinical and pathological characteristics, and recurrence rates of stage T1c versus T2a or T2b prostate cancer. J Urol 1999; 161:1525-9. [PMID: 10210388] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [MESH Headings] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/12/2023]
Abstract
PURPOSE We compare clinicopathological features, and cancer recurrence and survival rates in men with stage T1c versus T2a or T2b prostate cancer. MATERIALS AND METHODS From 1988 through 1998, 1 surgeon (W. J. C.) performed radical retropubic prostatectomy in 1,620 men with a mean age plus or minus standard deviation of 62.3 +/- 7 years. Clinical stage was T1c in 39%, T2a in 22% and T2b in 39% of patients. Patients were followed with semiannual prostate specific antigen (PSA) measurement and annual digital rectal examination. Serum total PSA greater than 0.3 ng./ml., histologically confirmed local tumor recurrence or distant metastases were considered evidence of cancer recurrence. Simple univariate statistics were used to compare clinical and pathological features by clinical stage, and multivariate Cox models were used to compare 5-year recurrence-free probabilities . The 5-year all cause and disease specific survival rates were calculated using Kaplan-Meier product limit estimates. RESULTS Mean patient age was younger for the clinical stage T1c group (61 years) than for the T2a (62 years) or T2b (64 years) group. Mean preoperative PSA and the percentage of patients with biopsy Gleason score 8 to 10 were more favorable for the T1c (8 ng./ml., 3%) and T2a (7, 5%) groups than for the T2b group (11, 6%). Cancerous surgical margins, seminal vesicle invasion and lymph node metastases were also less frequent in the T1c (20, 5 and 0.8%, respectively) and T2a (23, 5 and 0.3%) groups than in the T2b group (29, 11 and 1.8%). The 5-year recurrence-free survival rate was 85% for T1c, 83% for T2a and 72% for T2b cases. Multivariate analysis indicated a decreased risk of recurrence for the T1c group compared to the T2a and T2b groups. The 5-year disease specific survival rate was 100% for the T1c and T2a groups, and 97% for the T2b group. CONCLUSIONS Clinical and pathological features were similar for stages T1c and T2a, and different from stage T2b cancers. The 5-year recurrence-free survival was similar for T1c and T2a (log rank 0.89, p = 0.34), and higher than that for T2b (log rank 34.5, p <0.0001) cancers. However, controlling for all other prognostic factors on a Cox multivariate model, the risk of cancer recurrence was decreased for T1c compared to T2a and T2b disease. The detection of nonpalpable prostate cancer appears to be advantageous for intermediate-term cancer control.
Collapse
Affiliation(s)
- C G Ramos
- Department of Surgery, Washington University School of Medicine, St. Louis, Missouri, USA
| | | | | | | | | |
Collapse
|
9
|
Ramos CG, Carvalhal GF, Smith DS, Mager DE, Catalona WJ. Retrospective comparison of radical retropubic prostatectomy and 125iodine brachytherapy for localized prostate cancer. J Urol 1999; 161:1212-5. [PMID: 10081871] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [What about the content of this article? (0)] [Affiliation(s)] [Abstract] [MESH Headings] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/11/2023]
Abstract
PURPOSE Favorable results with 125iodine (I) brachytherapy have been reported in select patients with localized prostate cancer. We evaluate the results of radical prostatectomy in patients matched for similar pretreatment clinicopathological characteristics. MATERIALS AND METHODS From May 1983 to April 1998, 1 surgeon (W. J. C.) performed radical retropubic prostatectomy in 1,952 men (mean age plus or minus standard deviation 63+/-7 years), of whom 1,364 had Gleason score 6 or less on preoperative needle biopsy, a preoperative serum prostate specific antigen (PSA) value available and clinical stage T1 or T2 disease. We categorized all patients by preoperative Gleason score, preoperative PSA and clinical stage. For each Gleason score-by-PSA stratum we randomly selected by computer the number of men necessary to achieve the same overall distribution of clinical characteristics as in a series of patients treated with brachytherapy. All men were followed with semiannual PSA measurements and annual digital rectal examinations. Serum PSA greater than 0.3 ng/ml was considered evidence of cancer recurrence. Simple univariate statistics were used to compare clinical characteristics between series, and 7-year recurrence-free survival was estimated using Kaplan-Meier product limit estimates. To avoid a possible chance extreme result from 1 random sample we estimated 7-year recurrence-free survival in 5 computer selected random samples of our population. RESULTS Mean 7-year recurrence-free survival was 84% (95% confidence intervals 78 to 89) for the radical prostatectomy series compared to 79% (confidence intervals not provided) for the 125I brachytherapy series. CONCLUSIONS Radical prostatectomy yielded a proportionately but not statistically significant higher 7-year probability of nonprogression than 125I brachytherapy in patients with favorable clinicopathological characteristics. Comparisons are confounded by residual differences in clinicopathological features of tumors between groups and different treatment end points to determine outcomes. Further prospective, randomized clinical trials are required for valid comparisons.
Collapse
Affiliation(s)
- C G Ramos
- Department of Surgery, Washington University School of Medicine, St. Louis, Missouri, USA
| | | | | | | | | |
Collapse
|
10
|
Abstract
A retrospective investigation of squash-related injuries incurred at a private and a public club in New York was undertaken to gain insight into the incidence and nature of such injuries. Telephone interviews were conducted with 200 randomly selected individuals to obtain their entire injury history. Sixty-nine of the 155 squash players contacted sustained injuries during their participation. Some had multiple injuries. This resulted in an overall injury rate of 44.5%. Strains and sprains accounted for over one-third of the injuries. Nearly one-half of the injuries involved the lower extremity, with the ankle and lower leg being injured most often. Forty-seven percent of the injuries seen were considered disabling because the patients were out of action for more than two weeks after injury. Players over the age of 40 are at increased risk for injury, as are newcomers to the game. Increased injury rates were also associated with an increased frequency of play among females. Only rarely will an injured squash player become permanently impaired.
Collapse
|