A deep learning-based approach for localization of pedicle regions in preoperative CT scans

Postoperative diffusion weighted MRI and preoperative CT scan fusion for residual cholesteatoma localization

606 Background: To evaluate the ability of Aorta−Lesion−Attenuation−Difference (ALAD) to differentiate malignant renal tumors from renal oncocytomas. Methods: A retrospective review of preoperative CT scans and surgical pathology from robotic assisted partial nephrectomy specimens obtained by a single surgeon was performed. ALAD was calculated by measuring the difference in Hounsfield units (HU) between the aorta and the lesion of interest on the same image slice in the nephrographic phase on preoperative CT scan. The discriminative ability of ALAD to differentiate malignant pathology from oncocytoma was evaluated by sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under curve (AUC) using ROC analysis. Results: A total of 218 preoperative CT scans and corresponding pathology reports were reviewed. Pathology review revealed 22 oncocytomas (10.1%), 11 chromophobe RCC (5%), 37 papillary RCC (17%), and 148 clear cell RCC (67.9%). ALAD was able to differentiate malignant pathology from oncocytoma using a HU threshold of 24 with a sensitivity of 84%, specificity of 86%, PPV of 98%, and NPV of 33%. The AUC for malignant pathology versus oncocytoma was 0.86 (95% CI 0.77−0.96). Subgroup analysis showed that ALAD was able to differentiate chromophobe RCC from oncocytoma using a HU threshold of 24 with a sensitivity of 100%, specificity of 86%, PPV of 75%, and a NPV of 100%. The AUC for chromophobe RCC versus oncocytoma was 0.98 (95% CI 0.91−1.00). Conclusions: ALAD measurements based upon preoperative CT scans provide good discrimination between malignant renal tumors and oncocytomas, potentially decreasing the need for biopsy in certain patients. ALAD also discriminates well between chromophobe RCC and oncocytoma, which may aid in the management of patients with indeterminate diagnoses of oncocytic neoplasm on biopsy. Further validation of ALAD will be necessary prior to routine use in clinical practice.

Introduction:Cholesteatoma is traditionally diagnosed by otoscopic examination and treated by surgery. The necessity for imaging in an uncomplicated case is controversial. This study was planned to investigate the usefulness of a preoperative high-resolution computed tomography (HRCT) scan in depicting the status of middle ear structures in the presence of cholesteatoma and also to compare the correspondence between pre- and intraoperative CT findings in patients with cholesteatoma.Materials and Methods:This prospective descriptive study was performed from January 2009 to May 2011 in 36 patients with cholesteatoma who were referred to the Kashani and Al-Zahra Clinics of Otolaryngology. Preoperative high-resolution temporal bone CT scans (axial and coronal views) were carried out and compared with intraoperative findings.Results:Evaluation of 36 patients and their CT scans revealed excellent correlation for sigmoid plate erosion, widening of aditus, and erosion of scutum; good correlation for erosion of malleus and tegmen; moderate correlation for lateral canal fistula (LCF) and erosion of mastoid air cells; and poor correlation for facial nerve dehiscence (FND), incus, and stapes erosion. Conclusion: A preoperative CT scan may be helpful in relation to diagnosis and decision making for surgery in cases of cholesteatoma and ossicular erosion. The CT scan can accurately predict the extent of disease and is helpful for detection of lateral canal fistula, erosions of dural plate, and ossicular erosions. However it is not able to distinguish between cholesteatoma and mucosal disease, facial nerve dehiscency, incus, and stapes erosion.

Bone mineral density surrounding the screw thread predicts the risk of pedicle screw loosening.

Introduction: Preoperative risk stratification of kidney cancer patients is critical for treatment selection and improving mortality rates. Current clinical approaches such as cancer grading are limited by inter-reviewer variability, require invasive biopsy procedures, and might not capture the full extent of patient risk. We sought to evaluate a deep learning (DL) model to quantify patient risk and prognosticate overall survival in kidney cancer patients via CT scans. Methods: This retrospective study used preoperative contrast CT scans from 4 public multi-institutional collections with date of death/censoring available. A deep learning (DL) network architecture was developed to analyze input CT scans via a convolutional feature extractor and a Cox proportional hazards-based cost function. The DL model assigned a risk score to each patient, with the median risk used as a cutoff for classifying patients as high-risk or low-risk. Model performance for prognosticating overall survival (OS) was evaluated separately in training, internal validation, and external validation cohorts via Kaplan-Meier analysis in terms of concordance index and hazard ratios, with risk groups compared via Wilcoxon log-rank tests (α = 0.05). Results: In the training/internal validation cohort (N=418, 8 institutions) and external validation cohort (N=102, 1 institution), the median age was 60 and 59 years, the male/female ratios were 62%/38% and 75%/25%, with a median follow up of 2.8 and 3.1 years, respectively. The DL model yielded a training c-index of 0.66 (95% CI: 0.63-0.68), internal validation c-index of 0.73 (95% CI: 0.70-0.76) and an external validation c-index of 0.67 (95% CI: 0.64-0.70). Kaplan-Meier analysis of associated survival curves based on DL-determined high- and low-risk patients demonstrated statistically significant separation in all cohorts: training (p = 0.003, HR = 8.63, 95% CI: 8.28-8.98), internal validation (p = 0.027, HR = 4.89, 95% CI: 4.52-5.26), and external validation (p = 0.01, HR = 6.54, 95% CI: 6.18-6.90). Baseline clinicopathological characteristics were not significantly different between DL high and low risk patients in external validation (chi-square testing; ISUP grade p=0.375, age p=0.824, race p=0.567, and sex p=0.440). ISUP grade was also not a significant predictor of OS in internal validation (HR=1.57, p=0.211, 95% CI: 1.28-1.85) or external validation (p = 0.465, HR=0.16, 95% CI: 0.33-0.74). Conclusion: A deep learning model based on pre-operative CT scans was found to accurately prognosticate overall survival in kidney cancers in a multi-institutional setting, independent of clinicopathological factors. This model is being validated on larger cohorts, and in the context of targeted therapies. Citation Format: Brennan Flannery, Mohsen Hariri, Thomas DeSilvio, Amir Sadri, Jane Nguyen, Erick M. Remer, Smitha Krishnamurthi, Satish E. Viswanath. Deep learning based risk stratification of pre-operative CT scans is prognostic of overall survival in kidney cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7379.

Laryngeal Tumor Volume as a Predictor for Thyroid Cartilage Penetration

The objective of this study was to evaluate the role of regional lymph node (LN) metastasis detected on preoperative CT and/or 18F-fluoro-2-deoxyglucose-positron emission tomography (FDG-PET) scans in the prediction of early tumor recurrence after curative surgical resection of pancreatic ductal adenocarcinoma (PDAC). This retrospective study included 137 patients who underwent upfront surgery with R0 resection of PDAC between 2013 and 2016. Regional LN metastasis was identified using two criteria: positive findings for regional LN metastasis on either preoperative CT or FDG-PET scans (LNOR), or on both preoperative CT and FDG-PET scans (LNAND). A total of 55 patients had early tumor recurrence within 12 months after curative resection. Univariable and multivariable Cox proportional hazard regression analysis showed that preoperative carbohydrate antigen 19–9 (CA19-9) levels, preoperative locally advanced status, and regional LN metastasis (both LNOR and LNAND criteria) were significant risk factors for early recurrence. Positive LNOR and LNAND showed significantly poorer recurrence-free survival compared to negative regional LN metastasis groups (p = 0.048 and p = 0.020, respectively). Compared with the LNAND criteria, the LNOR criteria provided higher sensitivity (22.4% vs. 15.5%, p = 0.046) and a higher negative predictive value (61.9% vs. 59.8%, p = 0.046). The LNOR definition provided more sensitive and accurate performance in diagnosing preoperative regional LN metastasis.

<p class="abstract"><strong>Background:</strong> Sino nasal polyposis is a common condition often encountered by the otolaryngologists. Though plain radiographs are valuable in the pre-operative evaluation, the CT scan plays a major role in identifying the extent of disease and the anatomic variations accurately for the surgeon. This study evaluates the value of pre-operative CT scans by correlating the scan findings with intra-operative surgical findings.</p><p class="abstract"><strong>Methods:</strong> Total 50 adult patients who attended the ENT department from November 2018 to October 2020 with Sino nasal polyposis were selected for the study. All were subjected to CT scan and underwent Functional Endoscopic Sinus Surgery. The intra-operative findings were compared with the CT scan findings. </p><p class="abstract"><strong>Results:</strong> In our study, the pre-operative CT scan showed single sinus involvement in 20% and multiple sinus involvement in 80% of cases. Mucosal abnormality was reported in 86% and normal in 14% of cases. Osteomeatal obstruction was reported in 44% and normal in 56% of cases. Various anatomical abnormalities were noted in 100% of cases. There was a perfect positive co relation between CT scan findings and intra operative findings with regards to Osteomeatal occlusion and anatomical or mucosal abnormalities. </p><p class="abstract"><strong>Conclusions:</strong> The study done to assess the correlation between pre-operative CT scan findings and intra-operative findings of Sino nasal polyposis cases has shown a perfect positive correlation between the findings. Pre-operative CT scan is an invaluable tool for the surgeon for accurate diagnosis and treatment planning. </p>

Undiagnosed and undertreated posterior malleolus fractures lead to early ankle instability and arthritis. A preoperative CT scan could improve the management of those fractures. This study assessed the benefits of a systematic ankle CT scanner to diagnose and manage posterior malleolus fracture. A monocentric retrospective cohort study was conducted. Sixty consecutive patients with bimalleolar fractures were operated and underwent a preoperative CT scan. The mean age was 50.0years old (18.6years old) with a mean body mass index of 20.3 (kg/m2) (11.4kg/m2) and 71.7% (43/60) of women. The primary outcome was the rate of posterior malleolus fragment diagnosed on X-rays and on CT scan. Secondly, interobserver and interobserver's agreement were compared between conventional X-rays and CT scan. Thirty-five (58.3%) posterior fragment fractures were observed on X-rays and 53 (88.3%) on the preoperative CT scan (p < 0.01). The intraobserver reproducibility for X-rays was low (0.02 [- 0.23; 0.27]) and moderate for CT scan (0.45 [0.0; 0.84]). The interobserver reproducibility for X-rays was moderate (0.39 [0.15; 0.60]) and excellent for CT scan (0.78 [0.0; 1.0]). A wide proportion of bimalleolar fractures are associated with posterior malleolus fractures and undiagnosed with standard X-rays. We advocate a systematic preoperative CT scan in the management of bimalleolar fractures. Level IV, retrospective cohort study. 2218999v0, date of registration: 11/08/2020 (retrospectively registered).

Fracture reduction planning and guidance in orthopaedic trauma surgery via multi-body image registration.

Purpose:Optic nerve invasion is an important cause of mortality in retinoblastoma. We aimed correlate preoperative imaging and surgical histopathology findings in enucleated eyes with retinoblastoma to determine the efficacy of preoperative imaging in predicting optic nerve invasion in retinoblastoma.Methods:A retrospective review of records of all patients undergoing primary enucleation for retinoblastoma at a tertiary eyecare system between March 2013 and December 2017 with all patients having undergone preoperative imaging, either CT scan or MRI. Data was analyzed statistically to determine the correlation between preoperative CT scan/MRI and histopathology.Results:Totally, 97 eyes of 97 patients were included in the study who underwent primary enucleation for unilateral retinoblastoma. The average age at presentation was 27.8 months with the chief complaint being leukocoria in all the cases. 14 patients (14.43%) had evidence of optic nerve involvement in preoperative imaging. 30 patients had optic nerve invasion on histopathology (laminar and retrolaminar). Spearman's rank correlation test revealed a significant correlation between MRI findings and HP and an insignficant correlation between CT findings and HP. The CT scan had a sensitivity of 20%, specificity of 88.89%, 50% positive predictive value and 66.67% negative predictive value. MRI had a sensitivity of 40%, specificity of 93.55%, positive predictive value of 66.67% and a negative predictive value of 82.86%.Conclusion:MRI showed significant moderate correlation with surgical histopathology for predicting optic nerve invasion in retinoblastoma whereas CT shows insignificant correlation with HPE. Therefore, we recommend MRI for predicting optic nerve invasion in cases of retinoblastoma.

: Multidetector computed tomography (MDCT) is emerging as a powerful noninvasive diagnostic tool. The appropriate role of this technique in the preoperative evaluation of cardiovascular disease has yet to be fully defined. Atrial fibrillation is the most common sustained cardiac arrhythmia, and novel minimally invasive surgical techniques have been developed to treat this condition by electrically isolating the pulmonary veins. The ideal methodology to preoperatively evaluate these patients remains debatable. We hypothesized that 64-slice CT could significantly affect perioperative planning. : Thirty-six consecutive patients who consented to undergo minimally invasive pulmonary vein isolation at our institution underwent a preoperative 64-slice cardiac CT scan. All cardiac and noncardiac abnormalities were recorded, and modifications to the initial surgical plan were documented. : The mean patient age was 64.4 ± 11.9 years [26 men (72.2%), 17 with known coronary artery disease (47.2%)]. Preoperative CT scanning detected 12 patients with abnormal pulmonary venous anatomy (33.3%), 3 with left atrial thrombus (8.3%), and 17 with significant coronary artery disease (47.2%). Furthermore, 20 studies (55.6%) detected pulmonary abnormalities (including 11 nodules). Preoperative scanning significantly altered surgical planning in 10 cases (27.8%). Alterations in patient treatment included preoperative invasive angiography, conversion of the mini-maze to an open chest procedure, alteration of surgical approach, and postponement/cancellation. : Sixty-four-slice CT scanning is a safe, rapid, and accurate procedure with important ramifications for surgical planning. This methodology could become an alternative approach to screen preoperative cardiac surgical patients.

IntroductionPulmonary segmentectomy, when combined with hilar and mediastinal lymphadenectomy, is currently considered the gold standard treatment for early-stage lung tumors (NSCLC) smaller than 2 cm in diameter. The preoperative planning for segmentectomies usually includes a contrast-enhanced CT with 2D reconstructions (axial, coronary, and sagittal). Recent technological advances allow 3D (volume rendering) reconstructions of preoperative CT scans, intended to improve the surgeon’s understanding of the segmental anatomy.The study aims to investigate the added value of 3D reconstruction in enhancing the surgeon’s understanding of anatomical structures, thus facilitating surgical planning and improving oncological outcomes.Methods and analysisThis is a prospective, randomized, controlled study.Patients will be randomized into two groups:1. Group 2D: the preoperative workup for these patients will consist of a contrast-enhanced chest CT with two-dimensional (2D) reconstructions (axial, coronary, and sagittal);2. Group 3D: the preoperative workup for these patients will consist of a contrast-enhanced chest CT with two-dimensional (2D) reconstructions (axial, coronary, and sagittal) and a 3D reconstruction (volume rendering) of the same chest CT employing dedicated software.The primary endpoints will be negative margin (R0) resection rate, resection margin (staple line-to-tumor distance), and thoracotomy conversions.We will use Fisher’s exact test for binary outcomes and Mann–Whitney U test for continuous outcomes. For subgroup analyses, we will use regression. Multivariable analyses will be based on logistic regression for binary outcomes and linear regression for continuous outcomes.Ethics and disseminationThe protocol and the model informed consent forms have been reviewed and approved by the ethics committee (N.: 1–2023) concerning scientific content and compliance with applicable research and human subject regulations.A Subcommittee on Publications was established to review all publications and report its recommendations to the steering committee. The anonymized participant-level dataset and statistical code for generating the results will not be publicly available.Trial registrationThe protocol was registered at ClinicalTrials.gov (ID: NCT05716815; Prospective rAndomized sTudy efficaCy tHree-dimensional rEconstructions Segmentectomy - Full-Text View - ClinicalTrials.gov). Jan 19, 2023.

ObjectivesSegmentation of anatomical structures on dento-maxillo-facial (DMF) computed tomography (CT) or cone beam computed tomography (CBCT) scans is increasingly needed in digital dentistry. The main aim of this research was to propose and evaluate a novel open source tool called DentalSegmentator for fully automatic segmentation of five anatomical structures on DMF CT and CBCT scans: maxilla/upper skull, mandible, upper teeth, lower teeth, and the mandibular canal. MethodsA retrospective sample of 470 CT and CBCT scans was used as a training/validation set. The performance and generalizability of the tool was evaluated by comparing segmentations provided by experts and automatic segmentations in two hold-out test datasets: an internal dataset of 133 CT and CBCT scans acquired before orthognathic surgery and an external dataset of 123 CBCT scans randomly sampled from routine examinations in 5 institutions. ResultsThe mean overall results in the internal test dataset (n = 133) were a Dice similarity coefficient (DSC) of 92.2 ± 6.3 % and a normalised surface distance (NSD) of 98.2 ± 2.2 %. The mean overall results on the external test dataset (n = 123) were a DSC of 94.2 ± 7.4 % and a NSD of 98.4 ± 3.6 %. ConclusionsThe results obtained from this highly diverse dataset demonstrate that this tool can provide fully automatic and robust multiclass segmentation for DMF CT and CBCT scans. To encourage the clinical deployment of DentalSegmentator, the pre-trained nnU-Net model has been made publicly available along with an extension for the 3D Slicer software. Clinical significanceDentalSegmentator open source 3D Slicer extension provides a free, robust, and easy-to-use approach to obtaining patient-specific three-dimensional models from CT and CBCT scans. These models serve various purposes in a digital dentistry workflow, such as visualization, treatment planning, intervention, and follow-up.

Introduction: Magnetic resonance imaging (MRI) is the most common tool to examine glioblastoma. Preoperative MRI can be used to initial diagnosis, and surgery planning, while follow-up MRIs can be used to evaluate treatment responses, identify recurrency, and detect side effects. The follow-up MRIs are usually taken after the first-line therapy, such as maximal safe resection. In the past, radiologists manually segment the tumor regions from normal brain tissue on follow-up MRIs, which is time-consuming, error-prone, and challenging. Several deep-learning (DL) models have been developed utilizing preoperative images, but their performance has yet to be evaluated on follow-up MRIs. In this research, we built the largest follow-up MRI cohort (311 patients) to assess these DL models and their generalizability and performance on independent preoperative and follow-up images. We also made the first follow-up-based deep learning models for this specific task. Methods: All evaluation deep learning models (10 models) were trained by the Brain Tumor Segmentation challenge 2020 (BraTS’20) and evaluated by fifty pairs of preoperative and follow-up scans from our institution. The segmentation form our institution is evaluated by board certified radiologist. MRIs in the BraTS’20 dataset were all preoperative scans. After the evaluation, we randomly assigned 264 patients’ scans from our institution to the training dataset and 47 patients’ scans to the testing dataset. We compared three types of models in our follow-up deep learning model, including 1) UNet-3D, 2) UNet-3D+transfer-learning, and 3) UNet-3D+transfer-learning+baysian-learning. The benchmark for all models was the Dice similarity coefficient (DSC). DSC can measure the spatial overlap between model prediction and ground truth. The value of DSC is between zero to one. Zero means no overlap and one indicates complete overlap. Results: Our study demonstrates that the BraTS'20 trained models' performance decreased by 13.05% in independent preoperative MRI scans and 19.04% in follow-up MRI scans. The most significant mismatch regions were FLAIR hyperintense regions (3.68% drop in independent preoperative scans and 10.61% drop in independent follow-up scans) and Non-enhancing core (5.20% drop in independent preoperative scans and 11.99% drop in independent follow-up scans). Our best model can achieve the best DSC among three tumor regions compared to all evaluation models (FLAIR hyperintense regions: DSC 0.77 V.S. DSC 0.58; Enhancing tumor region: DSC 0.87 V.S. DSC 0.68; Non-enhancing tumor region: DSC 0.92 V.S. DSC 0.55). Conclusion: Maximal safe resection induced brain structure change, decreasing the performance of the preoperative-based DL model. Implementing a follow-up MRI-based segmentation model is essential to make accurate and generalizable results to address structural changes after maximal safe resection. Our follow-up DL model demonstrates the DSC score can be recovered. We commit to further developing the tool to assist radiologists in handling follow-up MRIs for glioblastoma patients. Citation Format: Kang Lin Hsieh, Tanjida Kabir, Luis Nunez-Rubiano, Yu-Chun Hsu, Yu Cai, Juan Rodriguez Quintero, Octavio Arevalo, Kangyi Zhao, Jackie Zhang, Jiguang Zhu Zhu, Roy Riascos, Xioaqian Jiang, Shayan Shams. A confident and operator-independent deep segmentation model to measure residual tumor volume in the follow-up MRIs for glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 2 (Clinical Trials and Late-Breaking Research); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(8_Suppl):Abstract nr LB067.