Cross Validation Through Two-Dimensional Solution Surface for Cost-Sensitive SVM.

Model selection plays an important role in cost-sensitive SVM (CS-SVM). It has been proven that the global minimum cross validation (CV) error can be efficiently computed based on the solution path for one parameter learning problems. However, it is a challenge to obtain the global minimum CV error for CS-SVM based on one-dimensional solution path and traditional grid search, because CS-SVM is with two regularization parameters. In this paper, we propose a solution and error surfaces based CV approach (CV-SES). More specifically, we first compute a two-dimensional solution surface for CS-SVM based on a bi-parameter space partition algorithm, which can fit solutions of CS-SVM for all values of both regularization parameters. Then, we compute a two-dimensional validation error surface for each CV fold, which can fit validation errors of CS-SVM for all values of both regularization parameters. Finally, we obtain the CV error surface by superposing K validation error surfaces, which can find the global minimum CV error of CS-SVM. Experiments are conducted on seven datasets for cost sensitive learning and on four datasets for imbalanced learning. Experimental results not only show that our proposed CV-SES has a better generalization ability than CS-SVM with various hybrids between grid search and solution path methods, and than recent proposed cost-sensitive hinge loss SVM with three-dimensional grid search, but also show that CV-SES uses less running time.

Preterm neonatal lateral ventricle volume from three-dimensional ultrasound is not strongly correlated to two-dimensional ultrasound measurements.

The aim of this study is to compare longitudinal two-dimensional (2-D) and three-dimensional (3-D) ultrasound (US) estimates of ventricle size in preterm neonates with posthemorrhagic ventricular dilatation (PHVD) using quantitative measurements of the lateral ventricles. Cranial 2-D US and 3-D US images were acquired from neonatal patients with diagnosed PHVD within 10 min of each other one to two times per week and analyzed offline. Ventricle index, anterior horn width, third ventricle width, and thalamo-occipital distance were measured on the 2-D images and ventricle volume (VV) was measured from 3-D US images. Changes in the measurements between successive image sets were also recorded. No strong correlations were found between VV and 2-D US measurements ([Formula: see text] between 0.69 and 0.36). Additionally, weak correlations were found between changes in 2-D US measurements and 3-D US VV ([Formula: see text] between 0.13 and 0.02). A trend was found between increasing 2-D US measurements and 3-D US-based VV, but this was not the case when comparing changes between 3-D US VV and 2-D US measurements. If 3-D US-based VV provides a more accurate estimate of ventricle size than 2-D US measurements, moderate-weak correlations with 3-D US suggest that monitoring preterm patients with PHVD using 2-D US measurements alone might not accurately represent whether the ventricles are progressively dilating. A volumetric measure (3-D US or MRI) could be used instead to more accurately represent changes.

Incremental Support Vector Learning for Ordinal Regression.

Support vector ordinal regression (SVOR) is a popular method to tackle ordinal regression problems. However, until now there were no effective algorithms proposed to address incremental SVOR learning due to the complicated formulations of SVOR. Recently, an interesting accurate on-line algorithm was proposed for training ν -support vector classification (ν-SVC), which can handle a quadratic formulation with a pair of equality constraints. In this paper, we first present a modified SVOR formulation based on a sum-of-margins strategy. The formulation has multiple constraints, and each constraint includes a mixture of an equality and an inequality. Then, we extend the accurate on-line ν-SVC algorithm to the modified formulation, and propose an effective incremental SVOR algorithm. The algorithm can handle a quadratic formulation with multiple constraints, where each constraint is constituted of an equality and an inequality. More importantly, it tackles the conflicts between the equality and inequality constraints. We also provide the finite convergence analysis for the algorithm. Numerical experiments on the several benchmark and real-world data sets show that the incremental algorithm can converge to the optimal solution in a finite number of steps, and is faster than the existing batch and incremental SVOR algorithms. Meanwhile, the modified formulation has better accuracy than the existing incremental SVOR algorithm, and is as accurate as the sum-of-margins based formulation of Shashua and Levin.

Regression Segmentation for M³ Spinal Images.

Clinical routine often requires to analyze spinal images of multiple anatomic structures in multiple anatomic planes from multiple imaging modalities (M(3)). Unfortunately, existing methods for segmenting spinal images are still limited to one specific structure, in one specific plane or from one specific modality (S(3)). In this paper, we propose a novel approach, Regression Segmentation, that is for the first time able to segment M(3) spinal images in one single unified framework. This approach formulates the segmentation task innovatively as a boundary regression problem: modeling a highly nonlinear mapping function from substantially diverse M(3) images directly to desired object boundaries. Leveraging the advancement of sparse kernel machines, regression segmentation is fulfilled by a multi-dimensional support vector regressor (MSVR) which operates in an implicit, high dimensional feature space where M(3) diversity and specificity can be systematically categorized, extracted, and handled. The proposed regression segmentation approach was thoroughly tested on images from 113 clinical subjects including both disc and vertebral structures, in both sagittal and axial planes, and from both MRI and CT modalities. The overall result reaches a high dice similarity index (DSI) 0.912 and a low boundary distance (BD) 0.928 mm. With our unified and expendable framework, an efficient clinical tool for M(3) spinal image segmentation can be easily achieved, and will substantially benefit the diagnosis and treatment of spinal diseases.

Evaluation of MRI-TRUS fusion versus cognitive registration accuracy for MRI-targeted, TRUS-guided prostate biopsy.

OBJECTIVE: The purpose of this article is to compare transrectal ultrasound (TRUS) biopsy accuracies of operators with different levels of prostate MRI experience using cognitive registration versus MRI-TRUS fusion to assess the preferred method of TRUS prostate biopsy for MRI-identified lesions. SUBJECTS AND METHODS; One hundred patients from a prospective prostate MRI-TRUS fusion biopsy study were reviewed to identify all patients with clinically significant prostate adenocarcinoma (PCA) detected on MRI-targeted biopsy. Twenty-five PCA tumors were incorporated into a validated TRUS prostate biopsy simulator. Three prostate biopsy experts, each with different levels of experience in prostate MRI and MRI-TRUS fusion biopsy, performed a total of 225 simulated targeted biopsies on the MRI lesions as well as regional biopsy targets. Simulated biopsies performed using cognitive registration with 2D TRUS and 3D TRUS were compared with biopsies performed under MRI-TRUS fusion. RESULTS: Two-dimensional and 3D TRUS sampled only 48% and 45% of clinically significant PCA MRI lesions, respectively, compared with 100% with MRI-TRUS fusion. Lesion sampling accuracy did not statistically significantly vary according to operator experience or tumor volume. MRI-TRUS fusion-naïve operators showed consistent errors in targeting of the apex, midgland, and anterior targets, suggesting that there is biased error in cognitive registration. The MRI-TRUS fusion expert correctly targeted the prostate apex; however, his midgland and anterior mistargeting was similar to that of the less-experienced operators. CONCLUSION: MRI-targeted TRUS-guided prostate biopsy using cognitive registration appears to be inferior to MRI-TRUS fusion, with fewer than 50% of clinically significant PCA lesions successfully sampled. No statistically significant difference in biopsy accuracy was seen according to operator experience with prostate MRI or MRI-TRUS fusion.

Non-invasive assessment of functionally relevant coronary artery stenoses with quantitative CT perfusion: preliminary clinical experiences.

OBJECTIVES: We developed a quantitative Dynamic Contrast-Enhanced CT (DCE-CT) technique for measuring Myocardial Perfusion Reserve (MPR) and Volume Reserve (MVR) and studied their relationship with coronary stenosis. METHODS: Twenty-six patients with Coronary Artery Disease (CAD) were recruited. Degree of stenosis in each coronary artery was classified from catheter-based angiograms as Non-Stenosed (NS, angiographically normal or mildly irregular), Moderately Stenosed (MS, 50-80% reduction in luminal diameter), Severely Stenosed (SS, >80%) and SS with Collaterals (SSC). DCE-CT at rest and after dipyridamole infusion was performed using 64-slice CT. Mid-diastolic heart images were corrected for beam hardening and analyzed using proprietary software to calculate Myocardial Blood Flow (MBF, in mL∙min(-1)∙100 g(-1)) and Blood Volume (MBV, in mL∙100 g(-1)) parametric maps. MPR and MVR in each coronary territory were calculated by dividing MBF and MBV after pharmacological stress by their respective baseline values. RESULTS: MPR and MVR in MS and SS territories were significantly lower than those of NS territories (p < 0.05 for all). Logistic regression analysis identified MPR∙MVR as the best predictor of ≥50% coronary lesion than MPR or MVR alone. CONCLUSIONS: DCE-CT imaging with quantitative CT perfusion analysis could be useful for detecting coronary stenoses that are functionally significant.

Daily use of sildenafil improves endothelial function in men with type 2 diabetes.

Diminished vascular endothelial function results in decreased vasodilator capacity and is associated with erectile dysfunction (ED) in patients afflicted with type 2 diabetes. The current study was designed to evaluate whether daily use of sildenafil could alter endothelial function and improve penile rigidity in a group of patients with diabetic ED. A double-blind, placebo-controlled, prospective trial was conducted with 24 men with type 2 diabetes who were randomized into 2 groups: one receiving daily sildenafil (50 mg, n = 12) and the other placebo (n = 12) for 10 weeks. Erectile function was captured subjectively using the International Index of Erectile Function (IIEF-5), and endothelial function was objectively monitored via brachial artery flow-mediated dilation. Among the placebo and sildenafil groups, there were no significant differences in average patient age, time from type 2 diabetes diagnosis, duration of ED, or baseline IIEF-5 scores. Past medical histories, including smoking, alcohol consumption, hypertension, and hyperlipidemia, were also similar. At the conclusion of the 10-week trial, patients who received daily sildenafil had significantly improved erectile rigidity as captured by IIEF-5 (P < .001) and increased endothelial function via brachial artery flow-mediated dilation (P < .01). Endothelial function in men with type 2 diabetes was enhanced with daily sildenafil. Improved erectile rigidity and enhanced vascular circulation was noted after 10 weeks of daily sildenafil use.

Graph cuts with invariant object-interaction priors: application to intervertebral disc segmentation.

This study investigates novel object-interaction priors for graph cut image segmentation with application to intervertebral disc delineation in magnetic resonance (MR) lumbar spine images. The algorithm optimizes an original cost function which constrains the solution with learned prior knowledge about the geometric interactions between different objects in the image. Based on a global measure of similarity between distributions, the proposed priors are intrinsically invariant with respect to translation and rotation. We further introduce a scale variable from which we derive an original fixed-point equation (FPE), thereby achieving scale-invariance with only few fast computations. The proposed priors relax the need of costly pose estimation (or registration) procedures and large training sets (we used a single subject for training), and can tolerate shape deformations, unlike template-based priors. Our formulation leads to an NP-hard problem which does not afford a form directly amenable to graph cut optimization. We proceeded to a relaxation of the problem via an auxiliary function, thereby obtaining a nearly real-time solution with few graph cuts. Quantitative evaluations over 60 intervertebral discs acquired from 10 subjects demonstrated that the proposed algorithm yields a high correlation with independent manual segmentations by an expert. We further demonstrate experimentally the invariance of the proposed geometric attributes. This supports the fact that a single subject is sufficient for training our algorithm, and confirms the relevance of the proposed priors to disc segmentation.

Transvaginal Doppler-guided uterine artery occlusion for the treatment of symptomatic fibroids: summary results from two pilot studies.

OBJECTIVE: To evaluate the feasibility, safety, and short-term efficacy of bilateral uterine artery occlusion, using a transvaginal Doppler-guided vascular clamp as a minimally invasive therapy for symptomatic uterine leiomyomas. METHODS: We conducted two prospective, non-randomized, phase I pilot studies (Canadian Task Force Classification II-2) at a university-affiliated teaching hospital. Between June 2004 and May 2005, 30 premenopausal women with symptomatic uterine leiomyomas underwent bilateral uterine artery occlusion using a transvaginal Doppler-guided vascular clamp. Bilateral uterine artery occlusion was performed for 5.8 +/- 1.4 hours in the first 17 patients (Group 1) and from 6 to 9 hours (mean 7.05 +/- 1.0 hours) in the latter 13 patients (Group 2). Outcome measures included dominant fibroid volume (cm(3)), uterine volume (cm(3)), and improvement of menorrhagia at one, three, and six months. RESULTS: Bilateral occlusion of the uterine arteries was achieved in all 30 patients. In Group 1, the Ruta Menorrhagia Severity Scores decreased from baseline by 16%, 22% and 39% at one, three, and six months respectively. The dominant fibroid (DF) and uterine volumes decreased by 24% and 16% respectively at six months. In Group 2, the Ruta scores changed from baseline by +3%, -24%, and -42% at one, three, and six months respectively. The DF and uterine volumes decreased by 29% and 16%, respectively at six months. CONCLUSION: Following bilateral uterine artery occlusion using a transvaginal Doppler clamp, the dominant fibroid volume decreased by an average of 24%, uterine volume decreased by 12%, and menorrhagia symptoms were reduced by up to 42%. Uterine artery occlusion may provide the gynaecologist with an alternative to uterine artery embolization (UAE). The system is simple, easy to apply, and short-term efficacy may be equivalent to UAE.

Repeat prostate biopsy accuracy: simulator-based comparison of two- and three-dimensional transrectal US modalities.

PURPOSE: To compare the accuracy of biopsy with two-dimensional (2D) transrectal ultrasonography (US) with that of biopsy with conventional three-dimensional (3D) transrectal US and biopsy with guided 3D transrectal US in the guidance of repeat prostate biopsy procedures in a prostate biopsy simulator. MATERIALS AND METHODS: The institutional review board approved this retrospective study. Five residents and five experts performed repeat biopsies with a biopsy simulator that contained the transrectal US prostate images of 10 patients who had undergone biopsy. Simulated repeat biopsies were performed with 2D transrectal US, conventional 3D transrectal US, and guided 3D transrectal US (an extension of 3D transrectal US that enables active display of biopsy targets). The modalities were compared on the basis of time per biopsy and how accurately simulated repeat biopsies could be guided to specific targets. The probability for successful biopsy of a repeat target was calculated for each modality. RESULTS: Guided 3D transrectal US was significantly (P < .01) more accurate for simulated biopsy of repeat targets than was 2D or 3D transrectal US, with a biopsy accuracy of 0.86 mm +/- 0.47 (standard deviation), 3.68 mm +/- 2.60, and 3.60 mm +/- 2.57, respectively. Experts had a 70% probability of sampling a prior biopsy target volume of 0.5 cm(3) with 2D transrectal US; however, the probability approached 100% with guided 3D transrectal US. Biopsy accuracy was not significantly different between experts and residents for any modality; however, experts were significantly (P < .05) faster than residents with each modality. CONCLUSION: Repeat biopsy of the prostate with 2D transrectal US has limited accuracy. Compared with 2D transrectal US, the biopsy accuracy of both experts and residents improved with guided 3D transrectal US but did not improve with conventional 3D transrectal US.

Reliability and comparison of acromion assessment techniques on X-ray and magnetic resonance imaging (reliability of acromion assessment techniques).

OBJECTIVE: To determine the reliability and correlation of plain radiography and magnetic resonance imaging (MRI) in the assessment of acromion morphology. MATERIALS AND METHODS: Acromion morphology was assessed using the lateral acromion angle (LAA) and the acromion-humeral interval (AHI). Thirty patients who had X-rays and MRI for impingement syndrome were included. Six blinded observers assessed the acromion morphology subjectively and objectively. RESULTS: Neither acromion assessment technique demonstrated a positive correlation (kappa and intraclass coefficient < 0.55) between X-ray and MRI. Both techniques were reliable (kappa and intraclass coefficient > 0.55) when measured objectively by experienced observers. CONCLUSION: The LAA and the AHI are both reliable acromion assessment techniques on X-ray and MRI when measured objectively and by experienced observers.

The spectrum of ulnar collateral ligament injuries as viewed on magnetic resonance imaging of the metacarpophalangeal joint of the thumb.

OBJECTIVE: To elucidate the spectrum of ulnar collateral ligament (UCL) injuries detectable by magnetic resonance imaging (MRI). METHODS: Twenty-one patients (12 male and 9 female, aged 14-62 years) with acute hyperabduction injuries of the first metacarpophalangeal joint underwent MRI for clinically suspected UCL injuries. All scans were performed in either a large-bore, 1.5-T imager or an experimental small-bore, 1.9-T imager. MRI findings and clinical evaluations of all patients and surgical reports of those who underwent surgery (n = 10) were reviewed and correlated retrospectively. RESULTS: A total of 6 patients demonstrated injuries that did not fall into previously described categories of UCL injuries and therefore illustrated the existence of a subclass of UCL injuries. We divided the MRI findings into 5 categories: Stener's lesions (n = 6), moderately displaced (> or = 3 mm) complete tears (n = 5), minimally displaced (< 3 mm) complete tears (n = 4), nondisplaced complete tears (n = 3) and partial tears (n = 3). None of the MRI scans demonstrated a normal UCL. Although sensitivity and specificity were not calculated, only 2 cases demonstrated discordance between the MRI results and surgical findings or clinical outcomes. CONCLUSION: There is a spectrum of UCL injuries that have not previously been described.

Intrauterine herpes simplex infection.

Neonatal herpes simplex virus (HSV) infection is usually acquired at birth. We present an infant with intrauterine HSV infection acquired after rupture of membranes. The infant was born prematurely to a woman with a first episode of genital herpetic infection in early pregnancy and rupture of membranes 11 days prior to delivery. The infant, at delivery had extensive brain damage, hemorrhage, and cystic encephalomalacia due to herpes infection. On 2nd day of life, Magnetic Resonance Imaging, suggested that the hemorrhage had occurred 7 to 14 days before the scan and the placental pathology was supportive of an ascending infection. The case of an ascending herpetic intrauterine infection after rupture of the membranes raises the question of prevention and suppressive antiviral therapy.