Search

Purpose To analyze the cost-effectiveness of radioembolization in the treatment of intrahepatic cholangiocarcinoma (ICC) using the Surveillance, Epidemiology, and End Results (SEER) Medicare cancer database. Materials and Methods Cost as measured by total treatment-related reimbursement in patients diagnosed with ICC who received chemotherapy alone or chemotherapy and yttrium-90 radioembolization was assessed in the SEER Medicare cancer database (1999–2012). Survival analysis was performed, and incremental cost-effectiveness ratios were generated. Results The study included 585 patients. Average age at diagnosis was 71 years (standard deviation: 9.9), and 52% of patients were male. Twelve percent of patients received chemotherapy with radioembolization (n = 72), and 88% of patients (n = 513) received only chemotherapy. Median survival was 1043 days (95% confidence interval [CI]: 894–1244) for chemotherapy plus radioembolization and 811 days (95% CI: 705–925) for chemotherapy alone (P = .02). Patients who received combination therapy were slightly younger (71 vs 69 years, P = .03). No significant differences were observed between treatment groups in age at treatment, sex, race, or city size. Multivariable analysis showed a hazard ratio for progression for combination therapy versus chemotherapy alone of 0.76 (95% CI: 0.59–0.97, P = .029). The incremental cost-effectiveness ratio, a measure of cost of each added year of life, was $50,058.65 per year (quartiles: $11,454.63, $52,763.28). Conclusions Combination therapy of ICC with chemotherapy and radioembolization is associated with higher median survival and can be a cost-effective treatment, with a median cost of $50,058.65 per additional year of survival.

Hypothesis Patient-specific 3D printed models are useful presurgical planning tools because they accurately represent the anatomy and drilling characteristics of the middle cranial fossa (MCF) approach to the internal auditory canal (IAC). Background The MCF approach to the IAC can be challenging due to variability of the bony architecture along the floor of the middle fossa. Patient-specific 3D printed models may enhance surgeon knowledge of a given patient's anatomy when preparing for MCF surgery. Methods Six temporal bone models were printed from photoacrylic resin based on CT data obtained from cadaveric specimens using a desktop stereolithography 3D printer. Critical structures to avoid injuring, the facial nerve and membranous labyrinth, were modeled as hollow cavities and filled with indicator paint. Two neurotologists each drilled three 3D printed models and the corresponding cadaveric specimens, and then completed a 41-item visual analog scale questionnaire (score range of each item: 0-10) to assess the model's accuracy, utility, and potential as a training tool. Results Drilling the model was favorably rated (median score 9.2; range 7.3-9.6) for its ability to provide surgeons with an accurate mental image of the corresponding cadaveric anatomy. Overall similarity of feel of drilling the model in comparison to human bone was moderate (median 7.6; range 6.6-9.0). Surgeons would use this model to prepare for future cases (median 9.4; range 5.1-9.9) and felt it had excellent utility for training purposes (median 9.3; range 8.4-9.9). Conclusions Patient-specific 3D printed models provide an anatomically accurate and favorable tool for preparing for MCF surgery.

Background Abnormalities in size and position of the acetabulum have been linked to both developmental dysplasia of the hip and femoroacetabular impingement. Owing to its 3-dimensional (3D) complexity, plain radiography and cross-sectional studies [computed tomography (CT) and magnetic resonance imaging] have limitations in their ability to capture the complexity of the acetabular 3D anatomy. The goal of the study was to use 3D computed tomography reconstructions to identify the acetabular lunate cartilage and measure its size at varying ages of development and between sexes. Methods Patients aged 10 to 18 years with asymptomatic hips and a CT pelvis for appendicitis were reviewed. Patients were stratified by sex and age: preadolescent (10 to 12), young adolescent (13 to 15), and old adolescent (16 to 18) in equal proportions. Materialise 3-matic was used to generate a 3D pelvic model, and the acetabular lunate cartilage surface area was calculated. The lunate cartilage was divided into anatomic segments: superior (11:00 to 1:00), anterior (1:00 to 4:00), and posterior (8:00 to 11:00). The femoral head surface area was calculated to control for patient size. Mixed effects models were generated predicting segment size where side was treated as a repeated measure. Absolute and relative (lunate cartilage to femoral head) models were generated. Results Sixty-two patients (124 hips) were included. Females showed a significant decrease in femoral head coverage as age increased overall and in the 3 subsegments. The majority of changes occurred between the preadolescent and young adolescent groups. Males did not show an overall change, but the superior and anterior anatomic subgroups showed a significant decrease in coverage between the young and old adolescent groups. Male lunate cartilages were absolutely, but not relatively, larger than females. No clinically significant side-to-side differences were noted. Conclusions The relative femoral head coverage by the acetabular lunate cartilage reduced with increasing age, suggesting the growth of the femoral head outpaces the acetabular lunate cartilage's growth. This was more prominent in females. This study has important implications for expected acetabular coverage changes in the latter aspects of pediatric and adolescent development. Level of evidence Level III-diagnostic study.

Background Primary hyperparathyroidism is caused by single gland disease in 85% of cases and by multiglandular disease in 15%. Our aim was to discover if the addition of computed tomography acquisition/fusion on a hybrid scanner to traditional dual-phase single-photon emission tomography improves localization accuracy. Methods A prospective database was queried for the perioperative data of patients with sporadic primary hyperparathyroidism who had initial exploration from 2006–2014 with ≥6 month follow-up to define anatomy. Prior to 2010, patients had single-photon emission tomography (n = 633); after 2010, they had single-photon emission tomography/computed tomography (n = 755). Results In 1,388 patients, the rates of single gland disease (P = .8), bilateral exploration (P = .4), and negative imaging (145 patients, P = .33) were equal between imaging cohorts. In 1,186 patients with single gland disease, the positive predictive value of single-photon emission tomography/computed tomography was somewhat greater (90% vs 85%) and the accuracy of single-photon emission tomography/computed tomography was superior (83% vs 77%, P = .02). In the 202 patients with multiglandular disease, 20% had negative imaging results with no difference by type (single-photon emission tomography/computed tomography 17%, single-photon emission tomography 23%, P = .3), but single-photon emission tomography/computed tomography was more accurate (36%) than single-photon emission tomography (22%, P = .04) in predicting multiglandular disease. Conclusion In a large cohort study of patients with sporadic primary hyperparathyroidism undergoing parathyroidectomy, positive single-photon emission tomography/computed tomography results provided more reliable operative guidance than single-photon emission tomography for both single gland disease and multiglandular disease.

BACKGROUND AND PURPOSE: Cognitive deficits are among the most commonly reported post-concussive symptoms, yet the underlying microstructural injury is poorly understood. Our aim was to discover white matter injury underlying reaction time in mild traumatic brain injury DTI by applying transport-based morphometry. MATERIALS AND METHODS: In this retrospective study, we performed DTI on 64 postconcussive patients (10–28 years of age; 69% male, 31% female) between January 2006 and March 2013. We measured the reaction time percentile by using Immediate Post-Concussion Assessment and Cognitive Testing. Using the 3D transport-based morphometry technique we developed, we mined fractional anisotropy maps to extract the common microstructural injury associated with reaction time percentile in an automated manner. Permutation testing established statistical significance of the extracted injuries. We visualized the physical substrate responsible for reaction time through inverse transport-based morphometry transformation. RESULTS: The direction in the transport space most correlated with reaction time was significant after correcting for covariates of age, sex, and time from injury (Pearson r = 0.44, P < .01). Inverting the computed direction using transport-based morphometry illustrates physical shifts in fractional anisotropy in the corpus callosum (increase) and within the optic radiations, corticospinal tracts, and anterior thalamic radiations (decrease) with declining reaction time. The observed shifts are consistent with biologic pathways underlying the visual-spatial interpretation and response-selection aspects of reaction time. CONCLUSIONS: Transport-based morphometry discovers complex white matter injury underlying postconcussive reaction time in an automated manner. The potential influences of edema and axonal loss are visualized in the visual-spatial interpretation and response-selection pathways. Transport-based morphometry can bridge the gap between brain microstructure and function in diseases in which the structural basis is unknown.

To determine whether pathologic colorectal tumor KRAS mutation status is correlated with progression-free survival (PFS) by imaging after selective internal radiation therapy with Yttrium-90 (SIRT Y90) for metastatic colorectal cancer in the liver (mCRC). This was an IRB approved, HIPAA compliant retrospective cohort study. Consecutive patients with unresectable mCRC with documented KRAS mutation status treated at a single center from 2002 to 2013 with SIRT Y90 were investigated. Treatment response was compared between KRAS wild-type (wt) and mutant (mut) using an anatomic tumor response criteria based on RECIST 1.0. Kaplan–Meier estimation and Cox regression analysis were used to measure progression-free survival (PFS) and to assess independent prognostic factors for PFS. 82 of 186 patients met review criteria. 33 (40.2%) patients were identified as KRAS mut. PFS was longer in KRAS wt (median 166 days [95% CI 96–258 days]) vs. mut (median 91 days [95% CI 79–104 days], p = 0.002). KRAS mut patients were 1.48 times more likely to progress at first follow-up imaging than wt (95% CI 1.06–2.08, p = 0.024). Univariate analysis identified high pre-SIRT Y90 INR, KRAS wt, any use of anti-EGFR therapy, and post-SIRT Y90 chemotherapy as prognostic factors for longer PFS. In multivariate analysis, only KRAS wt was an independent prognostic factor for longer PFS (RR: 1.80 [95% CI 1.08–2.99], p = 0.024). Longer PFS is associated with KRAS wt vs. mut following SIRT Y90.

To investigate survival outcomes following radioembolization with Yttrium-90 (Y90) for neuroendocrine tumor liver metastases (NETLMs). This study was designed to assess the efficacy of Y90 radioembolization and to evaluate lung shunt fraction (LSF) as a predictor for survival.A single-center, prospective study of 44 consecutive patients (median age: 58.5 years, 29.5 % male) diagnosed with pancreatic (52.3 %) or carcinoid (47.7 %) NETLMs from 2006 to 2012 who underwent Y90 radioembolization was performed. Patients' baseline characteristics, including LSF and median overall survival (OS) from first Y90 radioembolization, were recorded and compared between patients with high (≥10 %) and low (10 %) LSF. Baseline comparisons were performed using Fisher's exact tests for categorical and Mann-Whitney U test for continuous variables. Survival was calculated using the Kaplan-Meier method. Univariate (Wilcoxon rank-sum test) and multivariate analyses (Cox Proportional Hazard Model) for risk factor analysis were performed.There was no statistically significant difference in age, gender, race, tumor properties, or previous treatments between patients with high (n = 15) and low (n = 29) LSF. The median OS was 27.4 months (95 %CI 12.73-55.23), with 4.77 months (95 %CI 2.87-26.73) for high and 42.77 months (95 %CI 18.47-59.73) for low LSF (p = 0.003). Multivariate analysis identified high LSF (p = 0.001), total serum bilirubin1.2 mg (p = 0.016), and lack of pretreatment with octreotide (p = 0.01) as independent prognostic factors for poorer survival. Tumor type and total radiation dose did not predict survival.LSF ≥10 %, elevated bilirubin levels, and lack of pretreatment with octreotide were found to be independent prognostic factors for poorer survival in patients with NETLMs.

To assess radiation dose in common pediatric diagnostic fluoroscopy procedures and determine the efficacy of dose tracking and dose reduction training to reduce radiation use.Fluoroscopy time and radiation dose area product (DAP) were recorded for upper GIs (UGI), voiding cystourethrograms (VCUGs), and barium enemas (BEs) during a six-month period. The results were presented to radiologists followed by a 1-hour training session on radiation dose reduction methods. Data were recorded for an additional six months. DAP was normalized to fluoroscopy time, and Wilcoxon testing was used to assess for differences between groups.Data from 1,479 cases (945 pretraining and 530 post-training) from 9 radiologists were collected. No statistically significant difference was found in patient age, proportion of examination types, or fluoroscopy time between the pre- and post-training groups (P ≥ .1), with the exception of a small decrease in median fluoroscopy time for VCUGs (1.0 vs 0.9 minutes, P = .04). For all examination types, a statistically significant decrease was found in the median normalized DAP (P.05) between pre- and post-training groups. The median (quartiles) for pretraining and post-training normalized DAPs (μGy·m(2) per minute) were 14.36 (5.00, 38.95) and 6.67 (2.67, 17.09) for UGIs; 13.00 (5.34, 32.71) and 7.16 (2.73, 19.85) for VCUGs; and 33.14 (9.80, 85.26) and 17.55 (7.96, 46.31) for BEs.Radiation dose tracking with feedback, paired with dose reduction training, can reduce radiation dose during diagnostic pediatric fluoroscopic procedures by nearly 50%.

The aim of the study is to assess patient outcomes, complications, impact on rehospitalizations, and healthcare costs in patients with malignant ascites treated with tunneled catheters. A total of 84 patients with malignant ascites (mean age, 60 years) were treated with tunneled catheters. Patients with peritoneal carcinomatosis and malignant ascites treated with tunneled drain catheter placement over a 3-year period were studied. Overall survival from the time of ascites and catheter placement were stratified by primary cancer and analyzed using the Kaplan–Meier method. Complications were graded by the Common Terminology Criteria for Adverse Events v3.0 (CTCAE). The differences between pre- and post-catheter admissions, hospitalizations, and Emergency Department (ED) visits, as well as related inpatient expenses were compared using paired t tests. There were no significant differences in gender, age, or race between different primary cancer subgroups. One patient (1 %) developed bleeding (CTCAE-2). Four patients (5 %) developed local cellulitis (CTCAE-2). Three patients (4 %) had prolonged hospital stay (between 7 and 10 days) to manage ascites-related complications such as abdominal distention, discomfort, or pain. Comparison between pre- and post-catheter hospitalizations showed significantly lower admissions (−1.4/month, p

239 Background: Assessment of the clinical outcome and prognostic factors of patients receiving Y90 radioembolization to treat colorectal liver metastases. Methods: Multi-institution retrospective study of patients with CRLM who received Y-90 radioembolization after undergoing molecular testing was performed. Patients treated between 2008 and 2019 were included from 7 institutions within the United States and Europe. Patient demographics, systemic therapy regimens, tumor characteristics, laboratory values and overall survival were analyzed between patients with differing histopathologic and genomic status. Genomic profiling was obtained for KRAS, BRAF, PIK3CA, AKT, MEK, NRAS and MMR genes. Kaplan-Meier survival estimation and multivariate Cox regression were analyzed. Results: 445 patients treated with Y90 radioembolization fulfilled the inclusion criteria. Of the total cohort, 430 patients were available who had documented KRAS mutation status. Average age at diagnosis was found to be 58.4 years for all patients (57.6% male, 42.4% female). Decreased survival post Y90 treatment was shown in those patients with mutant KRAS. The median survival in patients with mutated KRAS was 5.7 months, as compared to 8.0 months in those patients with wild-type KRAS (p< 0.0001). Patient demographics including age, gender and race were not shown to be significant in overall survival post-Y90 treatment (p-values >0.05). The average number of systemic chemotherapy treatments prior to Y90 was 2.15 and was similar between KRAS groups (mutant vs wild type.) KRAS status persisted as a significant predictor of survival on multivariate analysis (p< 0.0001; HR 2.1 CI: 1.54-2.85). Conclusions: KRAS status is shown as a significant prognostic factor in patients with colorectal liver metastases receiving Y90 radioembolization.

Lumbar transforaminal epidural steroid injection is a common and effective tool for managing lumbar radicular pain, although accidental intravascular injection can rarely result in paralysis. The purpose of this study is to determine the safest needle tip position for computed tomography (CT)-guided lumbar transforaminal epidural steroid injections as determined by incidence of intravascular injection.Three radiologists, in consensus, reviewed procedural imaging for consecutive CT-fluoroscopic lumbar transforaminal epidural steroid injections performed during a 16-month period. Intravascular injections were identified and categorized by needle tip position, vessel type injected, intravascular injection volume and procedural phase containing the intravascular injection. Pearson chi-square and logistic regression testing were used to assess differences between groups, as appropriate.Intravascular injections occurred in 9% (52/606) of injections. The intravascular injection rate was significantly lower (P0.001) for extraforaminal needle position (0%, 0/109) compared to junctional (8%, 27/319) and foraminal (14%, 25/178) needle tip positions. Of the intravascular injections, 4% (2/52) were likely arterial, 35% (18/52) were likely venous, and 62% (32/52) were indeterminate for vessel type injected. 46% (24/52) of intravascular injections were large volume, 33% (17/52) were small volume, and 21% (11/52) were trace volume. 56% (29/52) of intravascular injections occurred with the contrast trial dose, 29% (15/52) with the steroid/analgesic cocktail, and 15% (8/52) with both.An extraforaminal needle position for CT-fluoroscopic lumbar transforaminal epidural steroid injections decreases the risk of intravascular injection and therefore may be safer than other needle tip positions.

Summary Objective To evaluate if quantitative joint space width (JSW) measurements from radiographs correlate with 4-year Knee injury and Osteoarthritis Outcome Scores (KOOS) and clinical performance measures. Method The study group consisted of 942 patients with symptomatic knee osteoarthritis (OA). 4-year outcomes for six measures (KOOS Pain, Symptom, Quality of Life, and Function scores, 20-m walk pace, and chair stand time) were used to create six multiple linear regression models. Primary predictors were baseline minimum JSW and 4-year change in JSW measured from fixed flexion radiographs. Age, gender, body mass index (BMI), race, knee alignment, and baseline measures of the outcomes of interest were covariates. Results Lower baseline minimum JSW and a greater decrease in 4-year JSW significantly correlated with worse 4-year KOOS Pain, Symptom, and Quality of Life. With all other factors constant, a 4.1, 4.8, and 5.6 mm lower baseline JSW correlated with a clinically significant eight-point drop in 4-year KOOS Pain, Symptom, and Quality of Life scores respectively. Additionally, a 3.5, 3.1, and 4.0 mm loss of JSW over 4 years correlated with a clinically significant eight-point drop in 4-year KOOS Pain, Symptom, and Quality of Life scores respectively. Conclusions Our results indicate quantitative radiographic JSW measurements correlate with 4-year clinical outcomes. Since patients with narrower JSW at the onset of study had lower KOOS scores at 4 years even after controlling for 4-year change in JSW and baseline KOOS scores, clinical outcomes in knee OA may be predetermined once the disease process begins. These findings suggest early treatment with disease modifying therapies may be necessary to influence outcomes.

To assess the agreement between the true non-contrast (TNC) attenuation values of intra-abdominal structures and attenuation values obtained on virtual-unenhanced (VUE) images based on rapid kVp-switching dual-energy CT. The effects of contrast phase and patient characteristics (e.g., BMI, hematocrit, hemoglobin content) on VUE values were also investigated.Ninety four patients who underwent triphasic abdominal CT (liver mass protocol, n = 47; pancreas mass protocol, n = 47) between August 2014 and May 2015 were retrospectively reviewed. Unenhanced series was performed using conventional single-energy mode at 120 kVp. Late arterial and venous phase post-contrast series were obtained utilizing rapid kVp-switching dual-energy CT technique. VUE images were processed off of arterial (VUE-art) and venous (VUE-ven) phase series. Attenuation values of liver, pancreas, kidneys, adrenal glands, muscle, subcutaneous fat, aorta, IVC, and main portal vein were recorded on TNC and VUE sets of images. Attenuation values were compared using univariate linear regression and Student two-tailed paired t test.There was excellent correlation between TNC, VUE-art, and VUE-ven attenuation values across all organs (p 0.0001). Paired Student t test, however, showed significant difference between TNC and VUE-art attenuation of kidneys, right adrenal gland, paraspinal muscle, and aorta. There was also significant difference between TNC and VUE-ven attenuation of left kidney. Percentage of cases which had10 HU difference between VUE and TNC for an individual was calculated which ranged between 13% (right kidney) and 42% (right adrenal gland).Although the correlation between VUE and TNC attenuation values was excellent and mean difference between TNC and VUE attenuation values was negligible (ranging between -5.94 HU for paraspinal muscles to 6.2 HU in aorta), intra-patient analysis showed a considerable number of cases which had10 HU difference between VUE and TNC. VUE-ven generally offered a better approximation of TNC values. Further optimization of post-processing algorithms might be necessary before complete replacement of TNC with VUE images.

Planning imaging performed during CT-guided procedures may occasionally contain important incidental findings. The purpose of this study was to identify and characterize by clinical relevance the extraspinal findings detected on planning imaging for CT-guided lumbar transforaminal epidural steroid injections (TFESIs).Four radiologists retrospectively evaluated the planning scout views and CT studies for 488 consecutive CT-guided lumbar TFESIs performed in 400 patients over a 1-year period. Incidental extraspinal findings were identified and used to characterize patients by the need for follow-up using the CT Colonography Reporting and Data System (C-RADS), a classification scheme originally developed to characterize incidental findings on CT colonography. Patients with C-RADS E4 findings have potentially important findings that should be communicated to the referring physician; patients with C-RADS E3 findings have findings that are likely unimportant, but workup may be indicated. All previously unknown C-RADS E3 and E4 findings discovered in the course of this research were reported to referring physicians for appropriate patient follow-up.Ten of 400 (2.5%) patients were classified as C-RADS E4; the most common C-RADS E4 finding was vascular aneurysm or stenosis (4/400, 1.0%). Thirteen of 400 (3.3%) patients were classified as C-RADS E3; the most common C-RADS E3 finding was hepatomegaly (4/400, 1.0%). Of 22 patients with C-RADS E3 and E4 findings unknown to clinicians, the finding for only one (4.5%) was communicated to clinicians at the time of the procedure.Clinically important incidental extraspinal findings were identified in 5.8% of patients on the planning imaging performed for CT-guided lumbar TFESIs. Communication of clinically important findings was poor (4.5%).

BACKGROUND AND PURPOSE: Cervical transforaminal epidural steroid injections are commonly performed for temporary pain relief or diagnostic presurgical planning in patients with cervical radiculopathy. Intravascular injection of steroids during the procedure can potentially result in cord infarct, stroke, and even death. CT-fluoroscopy allows excellent anatomic resolution and precise needle positioning. This study sought to determine the safest needle tip position during CT-guided cervical transforaminal epidural steroid injection as determined by the incidence of intravascular injection. MATERIALS AND METHODS: We retrospectively evaluated procedural imaging for consecutive single-site CT-fluoroscopic cervical transforaminal epidural steroid injection performed during a 13-month period. Intravascular injections were identified and classified by volume, procedure phase, vessel type, and needle tip position relative to the targeted neural foramen. ANOVA, Wilcoxon, or Pearson χ(2) testing was used to assess differences among groups as appropriate. RESULTS: Intravascular injections occurred in 49/201 (24%) procedures. Of the intravascular injections, 13/49 (27%) were large, 10/49 (20%) were small, and 26/49 (53%) were trace volume. Sixteen of 49 (33%) intravascular injections occurred with a trial contrast dose; 27/49 (55%), with a steroid/analgesic cocktail; and 6/49 (12%), with both. Twenty-seven of 49 (55%) intravascular injections were likely venous, 22/49 (45%) were indeterminate, and none were likely arterial. The intravascular injection rate was significantly lower (P < .001) for the extraforaminal needle position (8/82, 10%) compared with junctional (27/88, 31%) and foraminal (14/31, 45%) needle tip positions. CONCLUSIONS: An extraforaminal needle position for CT-guided cervical transforaminal epidural steroid injection decreases the risk of intravascular injection and therefore may be safer than other needle tip positions.

We present the case of an 80-year-old male with squamous cell carcinoma with bilobar hepatic metastases who underwent targeted Yttrium-90 radioembolization of the right hepatic lobe lesion. Subsequently, there was complete regression of the nontargeted, left hepatic lobe lesion. This may represent the first ever reported abscopal effect in radioembolization. The abscopal effect refers to the phenomenon of tumor response in nontargeted sites after targeted radiotherapy. In this article, we briefly review the immune-mediated mechanisms responsible for the abscopal effect.

BACKGROUND AND PURPOSE: Principal component analysis, a data-reduction algorithm, generates a set of principal components that are independent, linear combinations of the original dataset. Our study sought to use principal component analysis of fractional anisotropy maps to identify white matter injury patterns that correlate with posttraumatic headache after mild traumatic brain injury. MATERIALS AND METHODS: Diffusion tensor imaging and neurocognitive testing with the Immediate Post-Concussion Assessment and Cognitive Test were performed in 40 patients with mild traumatic brain injury and 24 without posttraumatic headache. Principal component analysis of coregistered fractional anisotropy maps was performed. Regression analysis of the major principal components was used to identify those correlated with posttraumatic headache. Finally, each principal component that correlated with posttraumatic headache was screened against other postconcussive symptoms and demographic factors. RESULTS: Principal component 4 (mean, 7.1 ± 10.3) correlated with the presence of posttraumatic headache in mild traumatic brain injury (odds ratio per SD, 2.32; 95% CI, 1.29–4.67; P = .01). Decreasing principal component 4 corresponded with decreased fractional anisotropy in the midsplenium and increased fractional anisotropy in the genu of the corpus callosum. Principal component 4 identified patients with posttraumatic headache with an area under the receiver operating characteristic curve of 0.73 and uniquely correlated with posttraumatic headache and no other postconcussive symptom or demographic factors. CONCLUSIONS: Principal component analysis can be an effective data-mining method to identify white matter injury patterns on DTI that correlate with clinically relevant symptoms in mild traumatic brain injury. A pattern of reduced fractional anisotropy in the splenium and increased fractional anisotropy in the genu of the corpus callosum identified by principal component analysis can help identify patients at risk for posttraumatic headache after mild traumatic brain injury.