Use of deep learning methods for hand fracture detection from plain hand radiographs.

BACKGROUND: Patients with hand trauma are usually examined in emergency departments of hospitals. Hand fractures are frequently observed in patients with hand trauma. Here, we aim to develop a computer-aided diagnosis (CAD) method to assist physicians in the diagnosis of hand fractures using deep learning methods. METHODS: In this study, Convolutional Neural Networks (CNN) were used and the transfer learning method was applied. There were 275 fractured wrists, 257 fractured phalanx, and 270 normal hand radiographs in the raw dataset. CNN, a deep learning method, were used in this study. In order to increase the performance of the model, transfer learning was applied with the pre-trained VGG-16, GoogLeNet, and ResNet-50 networks. RESULTS: The accuracy, sensitivity, specificity, and precision results in Group 1 (wrist fracture and normal hand) dataset were 93.3%, 96.8%, 90.3%, and 89.7%, respectively, with VGG-16, were 88.9%, 94.9%, 84.2%, and 82.4%, respectively, with Resnet-50, and were 88.1%, 90.6%, 85.9%, and 85.3%, respectively, with GoogLeNet. The accuracy, sensitivity, specificity, and precision results in Group 2 (phalanx fracture and normal hand) dataset were 84.0%, 84.1%, 83.8%, and 82.8%, respectively, with VGG-16, were 79.4%, 78.5%, 80.3%, and 79.7%, respectively, with Resnet-50, and were 81.7%, 81.3%, 82.1%, and 81.3%, respectively, with GoogLeNet. CONCLUSION: We achieved promising results in this CAD method, which we developed by applying methods such as transfer learning, data augmentation, which are state-of-the-art practices in deep learning applications. This CAD method can assist physicians working in the emergency departments of small hospitals when interpreting hand radiographs, especially when it is difficult to reach qualified colleagues, such as night shifts and weekends.

In-bore MRI-guided prostate biopsy in a patient group with PI-RADS 4 and 5 targets: A single center experience.

PURPOSE: To determine the diagnostic yield of magnetic resonance imaging (MRI) guided in-bore biopsy in patients with high likelihood multiparametric MRI (mpMRI) findings, regarding overall and clinically significant prostate cancer (csPCa) detection rates and concordance of biopsy and radical prostatectomy (RP) Gleason scores (GS). METHODS: This retrospective study consisted of 277 Prostate Imaging Reporting and Data System (PI-RADS) assessment category 4 and 5 targets in 246 patients (mean age, 65.7 years; median prostate specific antigen value, 7.75 ng/mL) who had undergone in-bore biopsy at our institution between 2012 and 2020. Eighty-one patients who underwent RP were eligible for the concordance analysis of biopsy and RP specimen GS. RESULTS: Overall PCa detection rates were 80.5 % per patient (198/246) and 78 % per target (216/277) and 83.5 % and 67.4 % in primary (biopsy naive) and secondary (at least one negative prior biopsy) settings. csPCa was found in 63 % overall, 66 % of patients (132/200) in the primary, and 50 % of patients (23/46) in the secondary biopsy settings (p < 0.001). The prostate cancer detection rate was 68 % and 92 % in PI-RADS 4 and 5, respectively (p < 0.001). In the radical prostatectomy subcohort, 27.2 % of patients were upgraded, 8.6 % of patients were downgraded from needle biopsy. Significant complications occurred in 1.2 % of patients. CONCLUSIONS: MRI-guided in-bore prostate biopsy has a high detection rate of csPCa in primary and secondary biopsy cohorts. Biopsy results were satisfactory in terms of the number of positive cores, cancer percentage in positive cores, and concordance of GS in needle biopsy and RP specimen.

The relationship between amount of extra-prostatic extension and length of capsular contact: performances from MR images and radical prostatectomy specimens

Backround/aim: In prostate cancer, extraprostatic extension (EPE) is an unfavorable prognostic factor, and the grade of EPE is correlated with the prognosis. This study aims to evaluate the utility of length of capsular contact (LCC ) in predicting the grade of EPE by correlating the measurements from MRI images and the measurements performed from radical prostatectomy specimens. Materials and methods: MR images and specimens of 110 tumors are analyzed retrospectively. The specimens are used as reference to validate the presence of EPE and to measure the ground truth LCC. MR images are evaluated by two radiologists to identify the presence of EPE and to predict the LCC indirectly. Reliability, accuracy, sensitivity, and specificity of the evaluations are analyzed in comparison with the findings obtained from the specimens. Results: In detection of EPE existence, the radiologists achieve almost the same performance (all AUCs = 0.73) with optimal cut-off values lead to moderate sensitivity and specificity pairs (For cut-off = 15.8 mm; Se = 0.69, Sp = 0.68 and for cut-off of 14.5 mm: Se = 0.77, Sp = 0.62). In distinguishing high-grade EPE from low-grade EPE, the radiologists accomplish very similar performances (AUCs = 0.73 and 0.72) Optimal thresholds of 20.0 mm and 18.5 mm for the readers retrospectively reveal medium sensitivity and specificity pairs (Se = 0.64, Sp = 0.67; Se = 0.64, Sp = 0.67). Conclusion: Consistent LCC estimates can be obtained from MR images providing a beneficial metric for detecting the existence of EPE and for discriminating the grades of EPE.

The role of T2-weighted images in assessing the grade of extraprostatic extension of the prostate carcinoma.

PURPOSE: Extraprostatic extension (EPE) is an unfavorable prognostic factor and the grade of EPE is also shown to be correlated with the prognosis of prostate cancer. The current study assessed the value of prostate magnetic resonance imaging (MRI) in measuring the radial distance (RD) of EPE and the role of T2 WI signs in predicting the grade of EPE. MATERIALS AND METHODS: A total of 110 patients who underwent prostate MRI before radical prostatectomy are enrolled in this retrospective study. Eighty-four patients have organ confined disease and the remaining twenty-six patients have EPE all verified by histopathology. Prostate MRI examinations were conducted with 3T MRI scanner and phased array coil with the following sequences: T2 WI, T1 WI, DCE, DWI with ADC mapping, and high b-value at b = 1500 s/mm2. The likelihood of EPE with 5-point Likert scale was assigned, several MRI features were extracted for each dominant tumor identified by using T2 WI. Tumors with Likert scales 4-5 were evaluated further to obtain MRI-based RD. The relationship between pathological and MRI-determined RD was tested. Univariate and multivariate logistic regression models were developed to detect the grade of pathological EPE. The inputs were among the 2 clinical parameters and 4 MRI features. RESULTS: There is a moderate correlation between pathological RD and MRI-determined RD (ρ = 0.45, P < 0.01). In univariate and multivariate models, MRI features and clinical parameters possess varying significance levels (univariate models; P = 0.048-0.788, multivariate models; P = 0.173-0.769). Multivariate models perform better than the univariate models by offering fair to good performances (AUC = 0.69-0.85). The multivariate model that employs the MRI features offers better performance than the model employs clinical parameters (AUC = 0.81 versus 0.69). CONCLUSION: Co-existence of T2 WI signs provide higher diagnostic value even than clinical parameters in predicting the grade of EPE. Combined use of clinical parameters and MRI features deliver slightly superior performance than MRI features alone.

Can Extraprostatic Extension Be Predicted by Tumor-Capsule Contact Length in Prostate Cancer? Relationship With International Society of Urological Pathology Grade Groups.

OBJECTIVE. The objective of our study was to evaluate the relationship between the tumor-capsule contact length, defined as tumor contact length (TCL), and extraprostatic extension (EPE) using the MRI-based TCL measurements and the real TCL measurements from pathology and to determine whether the International Society of Urological Pathology (ISUP) grade group of the tumors influenced this relationship. MATERIALS AND METHODS. In this retrospective study, we reviewed prostate multiparametric MRI (mpMRI) studies performed between 2012 and 2018 of 1576 patients and found that 134 patients also underwent radical prostatectomy (RP) after mpMRI. Finally, 86 patients with index lesions in contact with the prostate capsule in RP specimens were enrolled in the study. ROC analysis was used to evaluate the cutoff values of TCLs measured at pathology and TCLs measured on MRI in terms of EPE according to ISUP grade groups. RESULTS. There was no statistically significant cutoff value for pathology-based TCL measurements in individual ISUP grade groups and subgroups. Although not statistically significant, pathology-based TCL cutoff values decreased (from 21.0 to 11.0 mm) as ISUP grade group increased in terms of EPE positivity. When the relationship between MRI-based TCL measurements and EPE was considered, statistically significant cutoff values (range, 14.5-16.6 mm) could be determined in many groups and subgroups with low ISUP grades (sensitivity, 66.7-100%; specificity, 52.8-93.0%; p = 0.006-0.042). However, no statistically significant cutoff value was found for high ISUP grades. CONCLUSION. ISUP grade groups may have an effect on the TCL-EPE relationship. When the MRI-based TCL and EPE relationship is evaluated independent of ISUP grade group, a cutoff value around 15-16 mm may be usable to predict EPE.

Evaluation of the most optimal multiparametric magnetic resonance imaging sequence for determining pathological length of capsular contact.

OBJECTIVES: To assess the most optimal multi-parametric magnetic resonance imaging sequence (Mp-MRI) in determining pathological length of capsular contact (LCC) for the diagnosis of prostate cancer extraprostatic extension (EPE). METHODS: 105 patients with prostate cancer who underwent Mp-MRI of prostate prior to radical prostatectomy were enrolled in this retrospective study. LCC was determined from T2-weighted images (T2WI), Apparent Diffusion Coefficient (ADC) map, dynamic contrast-enhanced MRI (DCE-MRI) separately by two blinded radiologists. The LCCs in patients with and without EPE were compared with Mann Whitney-U test. The relationship between pathological LCC and the LCC that was measured from each Mp-MRI sequences were calculated by using Spearman test. The ability of all individual Mp-MRI sequences in determining pathological LCC was calculated by drawing receiver operator characteristic (ROC) curves. The diagnostic accuracy of LCC based on each MRI sequences for EPE diagnosis was also calculated with ROC curve analysis. RESULTS: The patients with EPE had longer median LCC than patients without EPE for each Mp-MRI sequences and for both readers. In addition, the LCC showed a broader overlapping between patients with and without EPE on ADC map (reader-1, p = 0.01; reader-2, p = 0.01) when compared with T2WI (reader-1, p = 0.002; reader-2, p = 0.001) and DCE-MRI (reader-1, p = 0.001; reader-2, p = 0.001). LCC based on DCE-MRI showed the strongest correlation with pathological LCC. The area under the curve (AUC) based on LCC was higher when using the DCE-MRI (reader-1: 0.874, p = 0.030; reader-2: 0.862, p = 0.02) than when using T2WI and ADC map in predicting pathological LCC for both readers. While the LCC based on ADC map showed poor diagnostic accuracy, LCC based on T2WI and DCE-MRI had fair diagnostic accuracy for EPE diagnosis. CONCLUSION: The contact between prostate tumor and capsule seems to be a useful and objective parameter for evaluating the EPE of prostate cancer with Mp-MRI. More specifically, LCC based on DCE-MRI has highest correlation with pathological LCC and has better ability to predict pathological LCC when compared with other Mp-MRI sequences. However, the performance of LCC based on T2WI and DCE-MRI was similar for EPE diagnosis. It seems measurement of LCC from DCE-MRI and measurement of LCC from T2WI does not show any difference in clinical EPE assessment.

Unilateral paraneoplastic optic disc edema and retinal periphlebitis in pineal germinoma.

PURPOSE: To describe a unilateral ocular paraneoplastic syndrome in pineal germinoma. OBSERVATIONS: A 24-year-old male presented with diplopia, excessive thirst, and frequent urination. Cranial MRI showed a mass in pineal gland. Dorsal midbrain syndrome signs were present. Examination showed optic disc edema and segmental retinal periphlebitis in right and normal fundus in left eye. Rheumatologic work-up was negative. Brain biopsy confirmed pineal germinoma. Retinal findings were attributed to paraneoplastic syndrome. Resolution of optic disc edema and retinal periphlebitis occurred following chemotherapy and focal irradiation. CONCLUSIONS: and importance: To date there are two published case reports on bilateral optic disc edema and retinal periphlebitis occurring as a paraneoplastic syndrome in pineal germinoma. This is the first report on a patient with pineal germinoma who had unilateral paraneoplastic involvement characterized by optic disc edema and retinal periphlebitis and who showed complete resolution of ocular disease after treatment of underlying tumor.

Evaluation of Peripheral Zone Prostate Cancer Aggressiveness Using the Ratio of Diffusion Tensor Imaging Measures.

Purpose: To evaluate the aggressiveness of peripheral zone prostate cancer by correlating the Gleason score (GS) with the ratio of the diffusion tensor imaging (DTI) measures. Materials and Methods: Forty-two peripheral zone prostate tumors were imaged using DTI. Regions of interest focusing on the center of tumor foci and noncancerous tissue were used to extract statistical measures of mean diffusivity (MD) and fractional anisotroy (FA). Measure ratio was calculated by dividing tumor measure by noncancerous tissue measure. Results: Strong correlations are observable between GS and MD measures while weak correlations are present between GS and FA measures. Minimum tumor MD (MDmin) and the ratio of minimum MD (rMDmin) show the same highest correlation with GS (both ρ = -0.73). Between GS ≤ 7 (3 + 4) and GS ≥ 7 (4 + 3), differences are significant for all MD measures but for some FA measures. MD measures perform better than FA measures in discriminating GS ≥ 7 (4 + 3). Conclusion: Ratios of MD measures can be used in evaluation of peripheral zone prostate cancer aggressiveness; however tumor MD measures alone perform similarly.

A combined VBM and DTI study of schizophrenia: bilateral decreased insula volume and cerebral white matter disintegrity corresponding to subinsular white matter projections unlinked to clinical symptomatology.

PURPOSE: Grey matter and white matter changes within the brain are well defined in schizophrenia. However, most studies focused on either grey matter changes or white matter integrity separately; only in limited number of studies these changes were interpreted in the same frame. In addition, the relationship of these findings with clinical variables is not clearly established. Here, we aimed to investigate the grey matter and white matter changes in schizophrenia patients and exhibit the relation of these imaging findings with clinical variables. METHODS: A total of 20 schizophrenia patients and 16 matched healthy controls underwent magnetic resonance imaging to investigate the grey matter and white matter alterations that occur in schizophrenia patients using voxel-based morphometry (VBM) and whole brain voxel-wise analysis of diffusion tensor imaging (DTI) parameters with SPM8, respectively. While the preprocessing steps of VBM were performed with the default parameters of VBM8 toolbox, the preprocessing steps of DTI were carried out using FSL. Additionally, VBM results were correlated with clinical variables. RESULTS: Bilateral insula showed decreased grey matter volume in schizophrenia patients compared with healthy controls (P < 0.01). The opposite contrast did not show a significant difference. Psychiatric scores, duration of illness, and age were not correlated with the decreased grey matter volume of insula in schizophrenia patients. DTI analysis revealed a significant increase in mean, radial, and axial diffusivity, mainly of the fibers of bilateral anterior thalamic radiation and superior longitudinal fasciculus with left predominance, which intersected with bilateral subinsular white matter (P < 0.05). CONCLUSION: Our findings suggest that insula may be the main affected brain region in schizophrenia, which is also well supported by the literature. Our results were independent of disease duration and schizophrenia symptoms. White matter alterations were observed within bilateral anterior thalamic radiation and superior longitudinal fasciculus that intersects with subinsular white matter. Studies with larger sample sizes and more detailed clinical assessments are required to understand the function of insula in the neurobiology of schizophrenia.

Rare case of glioblastoma multiforme located in posterior corpus callosum presenting with depressive symptoms and visual memory deficits.

Most of the primary brain tumours are located in the supratentorial region, and it is uncommon to see tumour growth on deep brain structures such as posterior corpus callosum (PCC). In addition, lesions in PCC are also difficult to recognise, because construction apraxia, visuospatial perception and attentional capacity impairment may be the only presenting symptoms. Here, we represent a rare case of gliobastoma multiforme located in PCC, which solely presents with depressive symptoms and visual memory deficits. Initial manifestations of primary brain tumours with psychiatric symptoms and memory disturbances, in addition to headaches and seizures, should be kept in mind.

Erratum to: Comparison of the trifecta outcomes of robotic and open nephron-sparing surgeries performed in the robotic era of a single institution.

[This corrects the article DOI: 10.1186/s40064-015-1274-2.].

Comparison of the trifecta outcomes of robotic and open nephron-sparing surgeries performed in the robotic era of a single institution.

PURPOSE: In this study we aimed to report a comparative analysis between open and robotic nephron sparing surgeries (NSS) from a single institutional database. METHODS: Patients who have undergone NSS during the robotic era of our institution were included in this study. Open (n = 74) and robotic (n = 59) groups were compared regarding trifecta outcome. Trifecta was defined as; warm ischemia time (WIT) <25 min, negative surgical margins and the absence of perioperative complications. RESULTS: A total of 57 (77 %) and 45 (76 %) patients in the open and robotic groups, respectively achieved the trifecta outcome. Overall trifecta rate was 77 % (n = 102/133). The only statistically significant difference between trifecta positive and trifecta negative patients was the length of hospitalization (LOH). Except LOH; none of the tested parameters were shown to be predictive of trifecta outcome on univariate and multivariate analyses. Concerning trifecta positive patients; those in the open surgery group had larger tumors with a higher degree of morphometric complexity and were hospitalized for a longer period of time. Additionally, operative duration was significantly higher in the robotic group. CONCLUSIONS: In our cohort, no significant difference in achieving the trifecta outcome was reported after open and robotic NSS. Length of hospitalization was the only parameter that differed significantly between trifecta positive and trifecta negative patients. Surgical approach was not a significant predictor of simultaneous achievement of trifecta outcomes. Irrespective of the trifecta definition; larger and more complicated tumors were handled via open NSS.

Multiparametric MRI guidance in first-time prostate biopsies: what is the real benefit?

PURPOSE: With the increased recognition of the capabilities of prostate multiparametric (mp) magnetic resonance imaging (MRI), attempts are being made to incorporate MRI into routine prostate biopsies. In this study, we aimed to analyze the diagnostic yield via cognitive fusion, transrectal ultrasound (TRUS)-guided, and in-bore MRI-guided biopsies in biopsy-naive patients with positive findings for prostate cancer screening. METHODS: Charts of 140 patients, who underwent transrectal prostate biopsy after the adaptation of mp-MRI into our routine clinical practice, were reviewed retrospectively. Patients with previous negative biopsies (n=24) and digital rectal examination findings suspicious for ≥cT3 prostate cancer (n=16) were excluded. T2-weighted imaging, diffusion-weighted imaging, and dynamic contrast-enhanced imaging were included in mp-MRI. Cognitive fusion biopsies were performed after a review of mp-MRI data, whereas TRUS-guided biopsies were performed blinded to MRI information. In-bore biopsies were conducted by means of real-time targeting under MRI guidance. RESULTS: Between January 2012 and February 2014, a total of 100 patients fulfilling the inclusion criteria underwent TRUS-guided (n=37), cognitive fusion (n=49), and in-bore (n=14) biopsies. Mean age, serum prostate specific antigen level, and prostate size did not differ significantly among the study groups. In TRUS-guided biopsy group, 51.3% were diagnosed with prostate cancer, while the same ratio was 55.1% and 71.4% in cognitive fusion and in-bore biopsy groups, respectively (P = 0.429). Clinically significant prostate cancer detection rate was 69.1%, 70.3%, and 90% in TRUS-guided, cognitive fusion, and in-bore biopsy groups, respectively (P = 0.31). According to histopathologic variables in the prostatectomy specimen, significant prostate cancer was detected in 85.7%, 93.3%, and 100% of patients in TRUS-guided, cognitive fusion, and in-bore biopsy groups, respectively. CONCLUSION: In the first set of transrectal prostate biopsies, mp-MRI guidance did not increase the diagnostic yield significantly.

Conspicuity of peripheral zone prostate cancer on computed diffusion-weighted imaging: comparison of cDWI1500, cDWI2000, and cDWI3000.

INTRODUCTION AND OBJECTIVE: Disadvantages associated with direct high b-value measurements may be avoided with use of computed diffusion-weighted imaging (cDWI). The purpose of this study is to assess the diagnostic performance of cDWI image sets calculated for high b-values of 1500, 2000, and 3000 s/mm(2). MATERIALS AND METHODS: Twenty-eight patients who underwent multiparametric MRI of the prostate and radical prostatectomy consecutively were enrolled in this retrospective study. Using a software developed at our institute, cDWI1500, cDWI2000, and cDWI3000 image sets were generated by fitting a monoexponential model. Index lesions on cDWI image sets were scored by two radiologists in consensus considering lesion conspicuity, suppression of background prostate tissue, distortion, image set preferability, and contrast ratio measurements were performed. RESULTS: Lesion detection rates are the same for computed b-values of 2000 and 3000 s/mm(2) and are better than b-values of 1500 s/mm(2). Best lesion conspicuity and best background prostate tissue suppression are provided by cDWI3000 image set. cDWI2000 image set provides the best zonal anatomical delineation and less distortion and was chosen as the most preferred image set. Average contrast ratio measured on these image sets shows almost a linear relation with the b-values. CONCLUSION: cDWI2000 image set with similar conspicuity and the same lesion detection rate, but better zonal anatomical delineation, and less distortion, was chosen as the preferable image set.

Final Gleason score prediction using discriminant analysis and support vector machine based on preoperative multiparametric MR imaging of prostate cancer at 3T.

OBJECTIVE: This study aimed at evaluating linear discriminant analysis (LDA) and support vector machine (SVM) classifiers for estimating final Gleason score preoperatively using multiparametric magnetic resonance imaging (mp-MRI) and clinical parameters. MATERIALS AND METHODS: Thirty-three patients who underwent mp-MRI on a 3T clinical MR scanner and radical prostatectomy were enrolled in this study. The input features for classifiers were age, the presence of a palpable prostate abnormality, prostate specific antigen (PSA) level, index lesion size, and Likert scales of T2 weighted MRI (T2w-MRI), diffusion weighted MRI (DW-MRI), and dynamic contrast enhanced MRI (DCE-MRI) estimated by an experienced radiologist. SVM based recursive feature elimination (SVM-RFE) was used for eliminating features. Principal component analysis (PCA) was applied for data uncorrelation. RESULTS: Using a standard PCA before final Gleason score classification resulted in mean sensitivities of 51.19% and 64.37% and mean specificities of 72.71% and 39.90% for LDA and SVM, respectively. Using a Gaussian kernel PCA resulted in mean sensitivities of 86.51% and 87.88% and mean specificities of 63.99% and 56.83% for LDA and SVM, respectively. CONCLUSION: SVM classifier resulted in a slightly higher sensitivity but a lower specificity than LDA method for final Gleason score prediction for prostate cancer for this limited patient population.