Clinical Computed Tomography Research Articles

Segmentation methods for quantifying X-ray Computed Tomography based biomarkers to assess hip fracture risk: a systematic literature review.

The success of using bone mineral density and/or FRAX to predict femoral osteoporotic fracture risk is modest since they do not account for mechanical determinants that affect bone fracture risk. Computed Tomography (CT)-based geometric, densitometric, and finite element-derived biomarkers have been developed and used as parameters for assessing fracture risk. However, to quantify these biomarkers, segmentation of CT data is needed. Doing this manually or semi-automatically is labor-intensive, preventing the adoption of these biomarkers into clinical practice. In recent years, fully automated methods for segmenting CT data have started to emerge. Quantifying the accuracy, robustness, reproducibility, and repeatability of these segmentation tools is of major importance for research and the potential translation of CT-based biomarkers into clinical practice. A comprehensive literature search was performed in PubMed up to the end of July 2024. Only segmentation methods that were quantitatively validated on human femurs and/or pelvises and on both clinical and non-clinical CT were included. The accuracy, robustness, reproducibility, and repeatability of these segmentation methods were investigated, reporting quantitatively the metrics used to evaluate these aspects of segmentation. The studies included were evaluated for the risk of, and sources of bias, that may affect the results reported. A total of 54 studies fulfilled the inclusion criteria. The analysis of the included papers showed that automatic segmentation methods led to accurate results, however, there may exist a need to standardize reporting of accuracy across studies. Few works investigated robustness to allow for detailed conclusions on this aspect. Finally, it seems that the bone segmentation field has only addressed the concept of reproducibility and repeatability to a very limited extent, which entails that most of the studies are at high risk of bias. Based on the studies analyzed, some recommendations for future studies are made for advancing the development of a standardized segmentation protocol. Moreover, standardized metrics are proposed to evaluate accuracy, robustness, reproducibility, and repeatability of segmentation methods, to ease comparison between different approaches.

Diagnostic Accuracy of Computed Tomography and Clinical Examination in the Assessment of Mandibular Invasion of Oral Squamous Cell Carcinomas

Introduction: Computed tomography (CT) scan precisely shows soft and hard tissues in the same test hereby determines the lesion extension, involvement of regional node as well as of bone. Current study was aimed to evaluate the efficacy of clinical examination and CT to assess mandibular invasion in oral squamous cell carcinoma. Materials and methods: This cross-sectional study was conducted at Dhaka Dental College Hospital from July, 2016 to July, 2017 among conveniently selected 35 patients of histologically confirmed squamous cell carcinoma which was close to the mandible. The patients underwent proper clinical examination. CT scan was performed; preoperative staging and treatment plan was formulated according to the status of bone invasion. After mandibulectomy, the resected specimens were sent for histopathology. The findings from clinical examination and CT were then correlated with the gold standard, postoperative histopathology. Results: Clinical examination accurately detected 22 cases to have bone invasion and 8 cases with no bone invasion. It also gave 2 false positive and 3 false negative results. On the other hand CT accurately detected 24 cases to have bone invasion and 9 cases with no bone invasion. It provided one false positive and one false negative result. However, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and accuracy of clinical examination were 88%, 80%, 91.67%, 72.73%, 85.71% respectively (p = 0.0002) and of CT were 96%, 90%, 96%, 90%, 94.28% respectively (p < 0.0001). Clinical examination and CT were found sensitive enough and have an acceptable range of specificity as primary investigative modalities. CT scan outperforms clinical examination in terms of sensitivity, specificity, NPV, and accuracy. Conclusion: CT scan imaging is the method of choice for planning treatment in advanced oral squamous cell carcinoma. This study reveals how a CT scan can provide additional diagnostic value to detect bone invasion. Update Dent. Coll. j: 2024; 14(2): 21-26

Medication of the cervical spinal nerves for the treatment of equine cervical radiculopathy

SummaryBackgroundSeveral perineural injection techniques have been described to medicate the cervical spinal nerves of horses with suspected radiculopathy due to lesions at the intervertebral foramen (IVF). The technique and outcome have never been described in live, standing horses with suspected clinical disease.ObjectivesTo assess the safety and efficacy of medication of the ventral rami of the cervical spinal nerves (VRCSNs) for the treatment of cervical radiculopathy in horses.Study designCase series.MethodsHorses that underwent ultrasound‐guided perineural injection of the VRCSNs following radiographic and/or computed tomography (CT) evaluation of the cervical spine were included. Perineural injection with corticosteroids was performed using a standardised technique under ultrasound guidance as previously described. Outcome data were collected by clinical record review and unstructured follow‐up telephone conversations.ResultsForty‐seven injections of the VRCSNs were performed in 12 horses with only two self‐limiting adverse reactions noted. All horses that underwent CT examination had at least one site of IVF narrowing with the foramina at the C6/C7 articulation most frequently affected. Nine horses improved following medication of the VRCSNs with eight returning to their previous levels of exercise. All seven horses that underwent CT of the cervical spine had an initial improvement in clinical signs. Six horses were in full work and still performing at their intended level of exercise at a median of 21.5 months post injection.Main limitationsLack of definitive diagnosis of spinal nerve compression without histopathology of the spinal nerves which is not possible in the live horse.ConclusionsMedication of the VRCSNs is a safe treatment option which may provide benefit to horses with suspected spinal nerve compression but should be reserved for horses that have undergone a detailed clinical examination and CT evaluation of the cervical spine.

Node Reporting and Data System Combined With Computed Tomography Radiomics Can Improve the Prediction of Nonenlarged Lymph Node Metastasis in Gastric Cancer.

To investigate the diagnostic performance of Node Reporting and Data System (Node-RADS) combined with computed tomography (CT) radiomics for assessing nonenlargement regional lymph nodes in gastric cancer (GC). Preoperative CT images were retrospectively collected from 376 pathologically confirmed of gastric adenocarcinoma from January 2019 to December 2023, with 605 lymph nodes included for analysis. They were divided into training (n = 362) and validation (n = 243) sets. Radiomics features were extracted from venous-phase, and the radiomics score was obtained. Clinical information, CT parameters, and Node-RADS classification were collected. A combined model was built using machine-learning approach and tested in validation set using receiver operating characteristic curve analysis. Further validation was conducted in different subgroups of lymph node short-axis diameter (SD) range. Node-RADS score, SD, maximum diameter of thickness of tumor, and radiomics were identified as the most predictive factors. The results demonstrated that the integrated model combining SD, maximum diameter of thickness of tumor, Node-RADS, and radiomics outperformed the model excluding radiomics, yielding an area under the receiver operating characteristic curve of 0.82 compared with 0.79, with a statistically significant difference (P < 0.001). Subgroup analysis based on different SDs of lymph nodes also revealed enhanced diagnostic accuracy when incorporating the radiomics score for the 4- to 7.9-mm subgroups, all P < 0.05. However, for the 8- to 9.9-mm subgroup, the combination of the radiomics did not significantly improve the prediction, with an area under the receiver operating characteristic curve of 0.85 versus 0.85, P = 0.877. The integration of radiomics scores with Node-RADS assessments significantly enhances the accuracy of lymph node metastasis evaluation for GC. This combined model is particularly effective for lymph nodes with smaller standard deviations, yielding a marked improvement in diagnostic precision. The findings of this study indicate that a composite model, which incorporates Node-RADS, radiomics features, and conventional parameters, may serve as an effective method for the assessment of nonenlarged lymph nodes in GC.

Comparison of Endoscopic Ultrasound and CT Scan in the Diagnosis of Esophageal Duplication Cysts

Background and AimEsophageal duplication cysts (EDCs) are rare congenital malformations, often discovered incidentally during endoscopy or on computed tomography (CT) scans. The role of endoscopic ultrasound (EUS) and CT scan in the diagnosis of these lesions and indications for surgical treatment are underreported. The aim of this study was to investigate these topics in a cohort of patients.Materials and MethodsBetween January 2001 and October 2020, 82 patients had a suspicion of esophageal duplication cyst on endoscopic ultrasound. Thirty four of these patients were referred for surgical enucleation of the lesion, but three patients were lost to follow-up. At the end, 31 patients, who underwent surgical treatment for their suspected EDC were included in this study. Clinical features, EUS findings, CT images, surgical treatment, and outcome were collected from hospital health records. CT images were re-evaluated by a chest radiologist. Type of surgery, surgical complications, and final histological diagnosis were reported.Results and ConclusionThe patients referred for surgery were younger (p = 0.0001) and had larger lesions (> 2 cm; p = 0.005) than the patients who had non-operative follow-up. From thirty-one operated patients, eighteen (58%) had post-operative histological diagnosis of duplication cyst. On EUS the final histological diagnosis was correct in 58% (18/31) of all the operated cases and on CT scan 57% (17/30). CT scan misdiagnosed three of the EDCs but found two leiomyomas correctly. None of these patients developed malignancy. According to this study, neither EUS without fine-needle biopsy nor CT scan alone can differentiate EDCs from other mediastinal masses.

Novel intravascular tantalum oxide-based contrast agent achieves improved vascular contrast enhancement and conspicuity compared to Iopamidol in an animal multiphase CT protocol

BackgroundTo assess thoracic vascular computed tomography (CT) contrast enhancement of a novel intravenous tantalum oxide nanoparticle contrast agent (carboxybetaine zwitterionic tantalum oxide, TaCZ) compared to a conventional iodinated contrast agent (Iopamidol) in a rabbit multiphase protocol.MethodsFive rabbits were scanned inside a human-torso-sized encasement on a clinical CT system at various scan delays after intravenous injection of 540 mg element (Ta or I) per kg of bodyweight of TaCZ or Iopamidol. Net contrast enhancement of various arteries and veins, as well as image noise, were measured. Randomized scan series were reviewed by three independent readers on a clinical workstation and assessed for vascular conspicuity and image artifacts on 5-point Likert scales.ResultsOverall, net vascular contrast enhancement achieved with TaCZ was superior to Iopamidol (p ≤ 0.036 with the exception of the inferior vena cava at 6 s (p = 0.131). Vascular contrast enhancement achieved with TaCZ at delays of 6 s, 40 s, and 75 s was superior to optimum achieved Iopamidol contrast enhancement at 6 s (p ≤ 0.036. Vascular conspicuity was higher for TaCZ in 269 of 300 (89.7%) arterial and 269 of 300 (89.7%) venous vessel assessments, respectively (p ≤ 0.005), with substantial inter-reader reliability (κ = 0.61; p < 0.001) and strong positive monotonic correlation between conspicuity scores and contrast enhancement measurements (ρ = 0.828; p < 0.001).ConclusionTaCZ provides absolute and relative contrast advantages compared to Iopamidol for improved visualization of thoracic arteries and veins in a multiphase CT protocol.Relevance statementThe tantalum-oxide nanoparticle is an experimental intravenous CT contrast agent with superior cardiovascular and venous contrast capacity per injected elemental mass in an animal model, providing improved maximum contrast enhancement and prolonged contrast conspicuity. Further translational research on promising high-Z and nanoparticle contrast agents is warranted.Key PointsThere have been no major advancements in intravenous CT contrast agents over decades.Iodinated CT contrast agents require optimal timing for angiography and phlebography.Tantalum-oxide demonstrated increased CT attenuation per elemental mass compared to Iopamidol.Nanoparticle contrast agent design facilitates prolonged vascular conspicuity.Graphical

Imaging performance of a LaBr3:Ce scintillation detector for photon counting x-ray computed tomography: Simulation study.

Photon counting detectors (PCDs) for x-ray computed tomography (CT) are the future of CT imaging. At present, semiconductor-based PCDs such as cadmium telluride (CdTe), cadmium zinc telluride, and silicon have been either used or investigated for clinical PCD CT. Unfortunately, all of them have the same major challenges, namely high cost and limited spectral signal-to-noise ratio (SNR). Recent studies showed that some high-quality scintillators, such as lanthanum bromide doped with cerium (LaBr3:Ce), are less expensive and almost as fast as CdTe. The objective of this study is to assess the performance of a LaBr3:Ce PCD for clinical x-ray CT. We performed Monte Carlo simulations and compared the performance of 3 mm thick LaBr3:Ce and 2 mm thick CdTe for PCD CT with x-rays at 120 kVp and 20-1000mA. The two PCDs were operated with either a threshold-subtract (TS) counting scheme or a direct energy binning (DB) counting scheme. The performance was assessed in terms of the accuracy of registered spectra, counting capability, and count-rate-dependent spectral imaging-task performance, for conventional CT imaging, water-bone material decomposition, and K-edge imaging with tungsten as the K-edge material. The performance for these imaging-tasks was quantified by nCRLB, that is, the Cramér-Rao lower bound on the variance of basis line-integral estimation, normalized by the corresponding value of CdTe at 20mA. The spectrum recorded by CdTe was distorted significantly due to charge sharing, whereas the spectra recorded by LaBr3:Ce better matched the incident spectrum. The dead time, estimated by fitting a paralyzable detector model to the count-rate curves, was 20.7, 15.0, 37.2, and 13.0ns for CdTe with TS, CdTe with DB, LaBr3:Ce with TS, and LaBr3:Ce with DB, respectively. Conventional CT imaging showed an adverse effect of reduced geometrical efficiency due to optical reflectors in LaBr3:Ce PCD. The nCRLBs (a lower value indicates a better SNR) for CdTe with TS, CdTe with DB, LaBr3:Ce with TS, LaBr3:Ce with DB, and the ideal PCD, were 1.00±0.01, 1.00±0.01, 1.18±0.02, 1.18±0.02, and 0.79±0.01, respectively, at 20mA. The nCRLBs for water-bone material decomposition, in the same order, were 1.00±0.02, 1.00±0.02, 0.85±0.02, 0.85±0.02, and 0.24±0.02, respectively, at 20mA; and 0.98±0.02, 0.98±0.02, 1.09±0.02, 0.83±0.02, and 0.24±0.02, respectively, at 1000mA. Finally, the nCRLBs for K-edge imaging, the most demanding task among the five, were 1.00±0.02, 1.00±0.02, 0.55±0.02, 0.55±0.02, and 0.13±0.02, respectively, at 20mA; and 2.45±0.02, 2.29±0.02, 3.12±0.02, 2.11±0.02, and 0.13±0.02, respectively, at 1,000mA. The Monte Carlo simulations showed that, compared to CdTe with either TS or DB, LaBr3:Ce with DB provided more accurate spectra, comparable or better counting capability, and superior spectral imaging-task performances, that is, water-bone material decomposition and K-edge imaging. CdTe had a better performance than LaBr3:Ce for the conventional CT imaging task due to its higher geometrical efficiency. LaBr3:Ce PCD with DB scheme may be an excellent alternative option for CdTe PCD.

Comparison of the Patient Demographics, High-Resolution Computed Tomography Features of the Pulmonary Ground-Glass Opacity and Its Diagnostic Value Analysis

Introduction Pulmonary ground-glass opacity (GGO) observed on computed tomography (CT) is widely regarded as a diagnostic feature of lung adenocarcinoma; however, a significant radiological predictive sign remains controversial. In this retrospective study, we conducted a comprehensive analysis of 206 patients with GGO to establish a correlation analysis model between CT images and diagnosis of GGO nodules. Methods Histopathologic specimens were obtained from 206 patients (130 women, 76 men; age range 24–77). The clinical data, pathologic findings, and thin-section CT features of solid, pure GGO (pGGO), and mixed GGO (mGGO) nodules were compared using rigorous statistical methods, such as t-test, Fisher’s exact test, or univariate logistic regression analysis. Results Significant differences were observed among the 3 groups in terms of patient demographics, including gender (P = 0.016), smoking history (P = 0.002), and nodule size (P = 0.002). Morphologic CT characteristics, such as the presence of spiculated sign, lobulated sign, vascular sign, bubble-lucency sign, or pleural retraction, exhibited significant differences among the solid nodule, pGGO, and mGGO groups. However, no significant differences were observed in terms of air-bronchogram sign. Notably, the incidence of malignancy was significantly higher in pGGO nodules (76.52%) than in solid (48.48%) and mGGO (73.86%) nodules. Patients with mGGO had up to 2.988 times higher hazard of malignant lesions than those with solid nodules [P = 0.036, hazard ratio (HR) = 2.988]. Similarly, the hazard of malignant lesions was 2.941 times higher in patients with pGGO than in those with solid nodules (P = 0.007, HR = 2.941). Conclusion Clinical, pathological, and thin-section CT features of solid, pGGO and mGGO nodules were found to be significantly different. Based on our comprehensive analysis, patients presenting with a mGGO or pGGO on CT scans were more likely to be diagnosed with lung cancer.

Tumour-pleura relationship on computed tomography (CT) provides effective risk stratification for peripheral pulmonary nodules with Lung Imaging Reporting and Data System (Lung-RADS) score of 4X.

Pulmonary nodules with Lung Imaging Reporting and Data System (Lung-RADS) 4X are of greater clinical significance, and accurate differentiation of pathological types and visceral pleural invasion (VPI) of Lung-RADS 4X peripheral pulmonary nodules before treatment can aid in stratification. This study set out to investigate whether the tumour-pleura relationship on computed tomography (CT) can provide effective risk stratification for peripheral pulmonary nodules with Lung-RADS 4X. This was a single institution, retrospective study of 482 consecutive patients with Lung-RADS score 4X, who were pathologically diagnosed with tuberculous granuloma and adenocarcinoma from January 2019 to December 2023. We assessed clinical factors (baseline characteristics and tumour markers) and CT findings. Univariate and multivariate logistic regression analyses were used to determine the classification of pulmonary nodules and predictors of VPI. Multivariate analysis revealed that gender [odds ratio (OR) =0.392; P<0.001], carcinoembryonic antigen (CEA) level (OR =8.331; P<0.001), type of nodules (OR =13.551 and 7.478; P<0.001 and P=0.016) and maximum base width of soft tissue component on the pleura side (OR =0.857; P=0.005) were significant independent factors for distinguishing tuberculous granuloma from adenocarcinoma. And the type of linear connection between lesion and pleura (OR =3.936; P<0.001), and the maximum base width of soft tissue components on the pleura side (OR =1.359; P=0.001) were correlated independently with VPI. The area under the curve (AUC) for predicting pulmonary nodules classification was 82.60% [95% confidence interval (CI): 78.85-86.35%), and the AUC for predicting VPI was 76.10% (95% CI: 69.83-82.38%). The tumour-pleura relationship will be helpful in further risk stratification for peripheral pulmonary nodules with a score of Lung-RADS 4X.

Chest CT features and risk factors for patients with Omicron variant pneumonia: a multicenter retrospective clinical study.

Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) Omicron variant infection has become widespread in China as a result of the alterations in epidemic control and prevention policies. We identified the clinical characteristics and lung computed tomography (CT) imaging characteristics of patients infected during the early stage of the Omicron BA.5 wave in Shanghai to provide a guide to the diagnosis, treatment, and prognosis of infection. Clinical information and lung CT imaging characteristics of patients with Omicron variant infection admitted to three designated hospitals in Shanghai from March to June 2022 were analyzed retrospectively. A total of 958 patients were included in the analysis. Among the patients, 169 (17.64%) had pneumonia confirmed by CT, of whom 70.41% (119/169) had lesions in < 10% of the lung area. Older age, unvaccinated status, and comorbid chronic lung disease, cerebrovascular disease, kidney disease, or Alzheimer`s disease were associated with poor prognosis. In patients with coronavirus disease 2019 (COVID-19) pneumonia, a large lesion size was associated with a poor prognosis. Age ≥ 65 years, unvaccinated status, fever > 5 days, and lymphocyte count < 0.5×109/L were risk factors for pneumonia. Age ≥ 65 years, unvaccinated status, fever > 5 days, and lymphocyte count < 0.5×109/L can be used to identify high-risk individuals who warrant a CT scan to screen for COVID-19 pneumonia, especially during the period of Omicron variant predominance. Concurrently, the importance of immunization should be emphasized to help people withstand the effects of Omicron variant infection.

Efficacy of automated machine learning models and feature engineering for diagnosis of equivocal appendicitis using clinical and computed tomography findings

This study evaluates the diagnostic efficacy of automated machine learning (AutoGluon) with automated feature engineering and selection (autofeat), focusing on clinical manifestations, and a model integrating both clinical manifestations and CT findings in adult patients with ambiguous computed tomography (CT) results for acute appendicitis (AA). This evaluation was compared with conventional single machine learning models such as logistic regression(LR) and established scoring systems such as the Adult Appendicitis Score(AAS) to address the gap in diagnostic approaches for uncertain AA cases. In this retrospective analysis of 303 adult patients with indeterminate CT findings, the cohort was divided into appendicitis (n = 115) and non-appendicitis (n = 188) groups. AutoGluon and autofeat were used for AA prediction. The AutoGluon-clinical model relied solely on clinical data, whereas the AutoGluon-clinical-CT model included both clinical and CT data. The area under the receiver operating characteristic curve (AUROC) and other metrics for the test dataset, namely accuracy, sensitivity, specificity, PPV, NPV, and F1 score, were used to compare AutoGluon models with single machine learning models and the AAS. The single ML models in this study were LR, LASSO regression, ridge regression, support vector machine, decision tree, random forest, and extreme gradient boosting. Feature importance values were extracted using the “feature_importance” attribute from AutoGluon. The AutoGluon-clinical model demonstrated an AUROC of 0.785 (95% CI 0.691–0.890), and the ridge regression model with only clinical data revealed an AUROC of 0.755 (95% CI 0.649–0.861). The AutoGluon-clinical-CT model (AUROC 0.886 with 95% CI 0.820–0.951) performed better than the ridge model using clinical and CT data (AUROC 0.852 with 95% CI 0.774–0.930, p = 0.029). A new feature, exp(-(duration from pain to CT)3 + rebound tenderness), was identified (importance = 0.049, p = 0.001). AutoML (AutoGluon) and autoFE (autofeat) enhanced the diagnosis of uncertain AA cases, particularly when combining CT and clinical findings. This study suggests the potential of integrating AutoML and autoFE in clinical settings to improve diagnostic strategies and patient outcomes and make more efficient use of healthcare resources. Moreover, this research supports further exploration of machine learning in diagnostic processes.

Noise suppression in photon-counting computed tomography using unsupervised Poisson flow generative models

Deep learning (DL) has proven to be important for computed tomography (CT) image denoising. However, such models are usually trained under supervision, requiring paired data that may be difficult to obtain in practice. Diffusion models offer unsupervised means of solving a wide range of inverse problems via posterior sampling. In particular, using the estimated unconditional score function of the prior distribution, obtained via unsupervised learning, one can sample from the desired posterior via hijacking and regularization. However, due to the iterative solvers used, the number of function evaluations (NFE) required may be orders of magnitudes larger than for single-step samplers. In this paper, we present a novel image denoising technique for photon-counting CT by extending the unsupervised approach to inverse problem solving to the case of Poisson flow generative models (PFGM)++. By hijacking and regularizing the sampling process we obtain a single-step sampler, that is NFE = 1. Our proposed method incorporates posterior sampling using diffusion models as a special case. We demonstrate that the added robustness afforded by the PFGM++ framework yields significant performance gains. Our results indicate competitive performance compared to popular supervised, including state-of-the-art diffusion-style models with NFE = 1 (consistency models), unsupervised, and non-DL-based image denoising techniques, on clinical low-dose CT data and clinical images from a prototype photon-counting CT system developed by GE HealthCare.

Clinical and computed tomography outcomes after mesh-enforced hiatoplasty and anterior hemi-fundoplication in large hiatal hernia repair.

The surgical technique in large hiatal hernia (HH) repair is controversially discussed and the outcome measures and follow-up schemes are highly heterogeneous. The aim of this study is to assess the true recurrence rate using computed tomography (CT) in patients with standardized large HH repair. Prospective single-center study investigating the outcome after dorsal, mesh-enforced large HH repair with anterior fundoplication. Endoscopy was performed after 3months and clinical follow-up and CT after 12months. Between 2012 and 2021, 100 consecutive patients with large HH were operated in the same technique. There were two reoperations within the first 90days for cephalad migration of the fundoplication. Endoscopic follow-up showed a correct position of the fundoplication and no relevant other pathologies in 99% of patients. Follow-up CT was performed in 100% of patients and revealed 6% of patients with a cephalad slippage, defined as migration of less than 3cm of the wrap, and 7% of patients with a recurrent hernia. One patient of each group underwent subsequent reoperation due to symptoms. There was no statistical correlation between abnormal radiological findings and clinical outcomes with 69.2% of patients being asymptomatic. Multivariate logistic regression did not show any prognostic factor for an unfavorable radiologic outcome. Ninety-four percent of patients rated their outcomes as excellent or good. Radiological follow-up after large HH repair using CT allows to detect slippage of the fundoplication wrap and small recurrences. Patients with unfavorable radiological outcomes rarely require operative revision but should be considered for further follow-up.

Artificial intelligence-measured nodule mass for determining the invasiveness of neoplastic ground glass nodules.

The nodule mass is an important indicator for evaluating the invasiveness of neoplastic ground-glass nodules (GGNs); however, the efficacy of nodule mass acquired by artificial intelligence (AI) has not been validated. This study thus aimed to determine the efficacy of nodule mass measured by AI in predicting the invasiveness of neoplastic GGNs. From May 2019 to September 2023, a retrospective study was conducted on 755 consecutive patients comprising 788 pathologically confirmed neoplastic GGNs, among which 259 were adenocarcinoma in situ (AIS), 282 minimally invasive adenocarcinoma (MIA), and 247 invasive adenocarcinoma (IAC). Nodule mass was quantified using AI software, and other computed tomography (CT) features were concurrently evaluated. Clinical data and CT features were compared using the Kruskal-Wallis test or Pearson chi-square test. The predictive efficacy of mass and CT features for evaluating invasive lesions (ILs) (MIAs and IACs) and IACs was analyzed and compared via receiver operating characteristic (ROC) analysis and the Delong test. ROC curve analysis revealed that the optimal cutoff value of mass for distinguishing ILs and AISs was 225.25 mg [area under the curve (AUC) 0.821; 95% confidence interval 0.792-0.847; sensitivity 64.27%; specificity 89.19%; P<0.001], and for differentiating IACs from AISs and MIAs, it was 390.4 mg (AUC 0.883; 95% confidence interval 0.858-0.904; sensitivity 80.57%; specificity 86.32%; P<0.001). The efficacy of nodule mass in distinguishing ILs and AISs was comparable to that of size (P=0.2162) and significantly superior to other CT features (each P value <0.001). Additionally, the ability of nodule mass to differentiate IACs from AISs and MIAs was significantly better than that of CT features (each P value <0.001). AI-based nodule mass analysis is an effective indicator for determining the invasiveness of neoplastic GGNs.

Clinical efficacy and computed tomography diagnostic value of bedaquiline‐containing regimens in the treatment of drug‐resistant pulmonary tuberculosis

AbstractObjectiveThis study investigated the clinical efficacy of bedaquiline‐containing regimens in the treatment of drug‐resistant pulmonary tuberculosis and the diagnostic value of computed tomography (CT).MethodsWe retrospectively analyzed the clinical diagnosis, treatment, and CT imaging data of patients with drug‐resistant pulmonary tuberculosis treated in Wenzhou Central Hospital from 1 January to 31 December 2022. According to whether the treatment regimen contained bedaquiline, the patients were divided into an observation group (bedaquiline tablets + background regimen) and a control group (background regimen). The clinical efficacy and pulmonary CT changes before and after treatment were analyzed in both groups.ResultsAfter 24 weeks of treatment, there was no statistically significant difference in the white blood cell count or concentrations of hemoglobin, alanine aminotransferase, serum albumin, or creatinine between the two groups (t = 0.71, 0.93, 0.05, 0.18, and 0.08, respectively; p &gt; 0.05). After 4, 8, and 12 weeks of treatment, there was no statistically significant difference in the sputum culture‐negative conversion rate between the two groups (χ2 = 2.67, 0.48, and 1.82, respectively; p &gt; 0.05). At 24 weeks of treatment, the sputum culture‐negative conversion rate in the observation group reached 100%, which was significantly higher than that in the control group (χ2 = 3.97, p &lt; 0.05). The effective absorption rates on chest imaging in the two groups of patients at 12 weeks were 83.33% and 57.89%, respectively. At 24 weeks of treatment, the effective absorption rates were 88.00% and 65.85% in the two groups, with a statistically significant difference (χ2 = 3.98; p &lt; 0.05). There were significant differences in cavity absorption at 24 weeks (χ2 = 4.33, p &lt; 0.05) and 48 weeks after treatment (χ2 = 10.63, p &lt; 0.05).ConclusionThe addition of bedaquiline to the background regimen improved the sputum culture‐negative conversion rate and chest imaging effective rate. Patients achieved good results at the end of the 24‐week treatment period.

Development of a multi-modal learning-based lymph node metastasis prediction model for lung cancer

PurposeThis study proposed a three-dimensional (3D) multi-modal learning-based model for the automated prediction and classification of lymph node metastasis in patients with non-small cell lung cancer (NSCLC) using computed tomography (CT) images and clinical information. MethodsWe utilized clinical information and CT image data from 4239 patients with NSCLC across multiple institutions. Four deep learning algorithm-based multi-modal models were constructed and evaluated for lymph node classification. To further enhance classification performance, a soft-voting ensemble technique was applied to integrate the outcomes of multiple multi-modal models. ResultsA comparison of the classification performance revealed that the multi-modal model, which integrated CT images and clinical information, outperformed the single-modal models. Among the four multi-modal models, the Xception model demonstrated the highest classification performance, with an area under the curve (AUC) of 0.756 for the internal test dataset and 0.736 for the external validation dataset. The ensemble model (SEResNet50_DenseNet121_Xception) exhibited even better performance, with an AUC of 0.762 for the internal test dataset and 0.751 for the external validation dataset, surpassing the multi-modal model's performance. ConclusionsIntegrating CT images and clinical information improved the performance of the lymph node metastasis prediction models in patients with NSCLC. The proposed 3D multi-modal lymph node prediction model can serve as an auxiliary tool for evaluating lymph node metastasis in patients with non-pretreated NSCLC, aiding in patient screening and treatment planning.

Exploring synthetic datasets for computer-aided detection: a case study using phantom scan data for enhanced lung nodule false positive reduction.

Synthetic datasets hold the potential to offer cost-effective alternatives to clinical data, ensuring privacy protections and potentially addressing biases in clinical data. We present a method leveraging such datasets to train a machine learning algorithm applied as part of a computer-aided detection (CADe) system. Our proposed approach utilizes clinically acquired computed tomography (CT) scans of a physical anthropomorphic phantom into which manufactured lesions were inserted to train a machine learning algorithm. We treated the training database obtained from the anthropomorphic phantom as a simplified representation of clinical data and increased the variability in this dataset using a set of randomized and parameterized augmentations. Furthermore, to mitigate the inherent differences between phantom and clinical datasets, we investigated adding unlabeled clinical data into the training pipeline. We apply our proposed method to the false positive reduction stage of a lung nodule CADe system in CT scans, in which regions of interest containing potential lesions are classified as nodule or non-nodule regions. Experimental results demonstrate the effectiveness of the proposed method; the system trained on labeled data from physical phantom scans and unlabeled clinical data achieves a sensitivity of 90% at eight false positives per scan. Furthermore, the experimental results demonstrate the benefit of the physical phantom in which the performance in terms of competitive performance metric increased by 6% when a training set consisting of 50 clinical CT scans was enlarged by the scans obtained from the physical phantom. The scalability of synthetic datasets can lead to improved CADe performance, particularly in scenarios in which the size of the labeled clinical data is limited or subject to inherent bias. Our proposed approach demonstrates an effective utilization of synthetic datasets for training machine learning algorithms.

Phenotyping COVID-19 respiratory failure in spontaneously breathing patients with AI on lung CT-scan

BackgroundAutomated analysis of lung computed tomography (CT) scans may help characterize subphenotypes of acute respiratory illness. We integrated lung CT features measured via deep learning with clinical and laboratory data in spontaneously breathing subjects to enhance the identification of COVID-19 subphenotypes.MethodsThis is a multicenter observational cohort study in spontaneously breathing patients with COVID-19 respiratory failure exposed to early lung CT within 7 days of admission. We explored lung CT images using deep learning approaches to quantitative and qualitative analyses; latent class analysis (LCA) by using clinical, laboratory and lung CT variables; regional differences between subphenotypes following 3D spatial trajectories.ResultsComplete datasets were available in 559 patients. LCA identified two subphenotypes (subphenotype 1 and 2). As compared with subphenotype 2 (n = 403), subphenotype 1 patients (n = 156) were older, had higher inflammatory biomarkers, and were more hypoxemic. Lungs in subphenotype 1 had a higher density gravitational gradient with a greater proportion of consolidated lungs as compared with subphenotype 2. In contrast, subphenotype 2 had a higher density submantellar–hilar gradient with a greater proportion of ground glass opacities as compared with subphenotype 1. Subphenotype 1 showed higher prevalence of comorbidities associated with endothelial dysfunction and higher 90-day mortality than subphenotype 2, even after adjustment for clinically meaningful variables.ConclusionsIntegrating lung-CT data in a LCA allowed us to identify two subphenotypes of COVID-19, with different clinical trajectories. These exploratory findings suggest a role of automated imaging characterization guided by machine learning in subphenotyping patients with respiratory failure.Trial registration: ClinicalTrials.gov Identifier: NCT04395482. Registration date: 19/05/2020.

Machine learning models can define clinically relevant bone density subgroups based on patient-specific calibrated computed tomography scans in patients undergoing reverse shoulder arthroplasty

Reduced bone density is recognized as a predictor for potential complications in reverse shoulder arthroplasty (RSA). While humeral and glenoid planning based on preoperative computed tomography (CT) scans assist in implant selection and position, reproducible methods for quantifying the patients' bone density are currently not available. The purpose of this study was to perform bone density analyses including patient specific calibration in an RSA cohort based on preoperative CT imaging. It was hypothesized that preoperative CT bone density measures would provide objective quantification of the patients' humeral bone quality. This study consisted of three parts, (1) analysis of a patient-specific calibration method in cadaveric CT scans, (2) retrospective application in a clinical RSA cohort, and (3) clustering and classification with machine learning models. Forty cadaveric shoulders were scanned in a clinical CT and compared regarding calibration with density phantoms, air muscle, and fat (patient-specific) or standard Hounsfield unit. Post-scan patient-specific calibration was used to improve the extraction of three-dimensional regions of interest for retrospective bone density analysis in a clinical RSA cohort (n=345). Machine learning models were used to improve the clustering (Hierarchical Ward) and classification (Support Vector Machine (SVM)) of low bone densities in the respective patients. The patient-specific calibration method demonstrated improved accuracy with excellent intraclass correlation coefficients (ICC) for cylindrical cancellous bone densities (ICC>0.75). Clustering partitioned the training data set into a high-density subgroup consisting of 96 patients and a low-density subgroup consisting of 146 patients, showing significant differences between these groups. The SVM showed optimized prediction accuracy of low and high bone densities compared to conventional statistics in the training (accuracy=91.2%; AUC=0.967) and testing (accuracy=90.5 %; AUC=0.958) data set. Preoperative CT scans can be used to quantify the proximal humeral bone quality in patients undergoing RSA. The use of machine learning models and patient-specific calibration on bone mineral density demonstrated that multiple 3D bone density scores improved the accuracy of objective preoperative bone quality assessment. The trained model could provide preoperative information to surgeons treating patients with potentially poor bone quality.

Automated segmentation of clinical CT scans of the cochlea and analysis of the cochlea's vertical profile

PurposeKnowledge of the cochlear anatomy in individual patients is helpful for improving electrode selection and placement during cochlear implantation, as well as in surgical planning. The aim of this study was to develop a model-free automated segmentation algorithm to obtain 3D surfaces from clinical computed tomography (CT) scans that describe an individual's cochlear anatomy and can be used to quantitatively analyze the cochlea's vertical trajectory. MethodsClinical CT scans were re-oriented and re-sliced to obtain mid-modiolar slices. Using these slices, we segmented the cross-section of the cochlea. Results3D surfaces were obtained for the first 1.5 turns of 648 cochleae. Validation of our algorithm against the manually segmented ground truth obtained from 8 micro-CT scans showed good agreement, with 90 % area overlap and an average distance of 0.11 mm between the segmentation contours. The average cochlear duct length for the basal turn was 16.1 mm along the central path and 22.4 mm along the outer wall. The use of an intrinsic, observer-independent coordinate system and principal component analysis enabled unambiguous quantitative evaluation of the vertical trajectory of the cochlea, revealing only a weak correlation between the symmetry of the commonly used basal turn diameters (B-ratio of A and B diameters) and the profile of the vertical trajectory. ConclusionA model-free segmentation algorithm can achieve similar accuracy as previously published methods relying on statistical shapes. Quantitative analysis of the vertical trajectory can replace the categorization into rollercoaster, sloping, or intermediate vertical trajectory types.