Large-scale pancreatic cancer detection via non-contrast CT and deep learning

Nature Medicine, Год журнала: 2023, Номер 29(12), С. 3033 - 3043

Опубликована: Ноя. 20, 2023

Pancreatic ductal adenocarcinoma (PDAC), the most deadly solid malignancy, is typically detected late and at an inoperable stage. Early or incidental detection associated with prolonged survival, but screening asymptomatic individuals for PDAC using a single test remains unfeasible due to low prevalence potential harms of false positives. Non-contrast computed tomography (CT), routinely performed clinical indications, offers large-scale screening, however, identification non-contrast CT has long been considered impossible. Here, we develop deep learning approach, pancreatic cancer artificial intelligence (PANDA), that can detect classify lesions high accuracy via CT. PANDA trained on dataset 3,208 patients from center. achieves area under receiver operating characteristic curve (AUC) 0.986-0.996 lesion in multicenter validation involving 6,239 across 10 centers, outperforms mean radiologist performance by 34.1% sensitivity 6.3% specificity identification, 92.9% 99.9% real-world multi-scenario consisting 20,530 consecutive patients. Notably, utilized shows non-inferiority radiology reports (using contrast-enhanced CT) differentiation common subtypes. could potentially serve as new tool screening.

Язык: Английский

---

Opportunistic Screening: Radiology Scientific Expert Panel

Radiology, Год журнала: 2023, Номер 307(5)

Опубликована: Май 23, 2023

Radiologic tests often contain rich imaging data not relevant to the clinical indication. Opportunistic screening refers practice of systematically leveraging these incidental findings. Although opportunistic can apply modalities such as conventional radiography, US, and MRI, most attention date has focused on body CT by using artificial intelligence (AI)-assisted methods. Body represents an ideal high-volume modality whereby a quantitative assessment tissue composition (eg, bone, muscle, fat, vascular calcium) provide valuable risk stratification help detect unsuspected presymptomatic disease. The emergence "explainable" AI algorithms that fully automate measurements could eventually lead their routine use. Potential barriers widespread implementation include need for buy-in from radiologists, referring providers, patients. Standardization acquiring reporting measures is needed, in addition expanded normative according age, sex, race ethnicity. Regulatory reimbursement hurdles are insurmountable but pose substantial challenges commercialization Through demonstration improved population health outcomes cost-effectiveness, CT-based should be attractive both payers care systems value-based models mature. If highly successful, justify standalone "intended" screening.

Язык: Английский

---

Environmental Sustainability and AI in Radiology: A Double-Edged Sword

Radiology, Год журнала: 2024, Номер 310(2)

Опубликована: Фев. 1, 2024

According to the World Health Organization, climate change is single biggest health threat facing humanity. The global care system, including medical imaging, must manage effects of while at same time addressing large amount greenhouse gas (GHG) emissions generated in delivery care. Data centers and computational efforts are increasingly contributors GHG radiology. This due explosive increase big data artificial intelligence (AI) applications that have resulted energy requirements for developing deploying AI models. However, also has potential improve environmental sustainability imaging. For example, use can shorten MRI scan times with accelerated acquisition times, scheduling efficiency scanners, optimize decision-support tools reduce low-value purpose this

Язык: Английский

---

AI Body Composition in Lung Cancer Screening: Added Value Beyond Lung Cancer Detection

Radiology, Год журнала: 2023, Номер 308(1)

Опубликована: Июль 1, 2023

Background An artificial intelligence (AI) algorithm has been developed for fully automated body composition assessment of lung cancer screening noncontrast low-dose CT the chest (LDCT) scans, but utility these measurements in disease risk prediction models not assessed. Purpose To evaluate added value CT-based AI-derived incidence, death, cardiovascular (CVD) and all-cause mortality National Lung Screening Trial (NLST). Materials Methods In this secondary analysis NLST, measurements, including area attenuation attributes skeletal muscle subcutaneous adipose tissue, were derived from baseline LDCT examinations by using a previously AI algorithm. The was assessed with sex- cause-specific Cox proportional hazards without predicting CVD mortality. Models adjusted confounding variables age; mass index; quantitative emphysema; coronary artery calcification; history diabetes, heart disease, hypertension, stroke; other PLCOM2012 factors. Goodness-of-fit improvements likelihood ratio test. Results Among 20 768 included participants (median age, 61 years [IQR, 57–65 years]; 12 317 men), 865 diagnosed 4180 died during follow-up. Including improved death (male participants: χ2 = 23.09, P < .001; female 15.04, .002), (males: 69.94, females: 16.60, .001), 248.13, 94.54, incidence 2.53, .11; 1.73, .19). Conclusion automatically predictive NLST. Clinical trial registration no. NCT00047385 © RSNA, 2023 Supplemental material is available article. See also editorial Fintelmann issue.

Язык: Английский

---

Biological age model using explainable automated CT-based cardiometabolic biomarkers for phenotypic prediction of longevity

Nature Communications, Год журнала: 2025, Номер 16(1)

Опубликована: Фев. 7, 2025

We derive and test a CT-based biological age model for predicting longevity, using an automated pipeline of explainable AI algorithms that quantifies skeletal muscle, abdominal fat, aortic calcification, bone density, solid organs. apply these tools to CT scans from 123,281 adults (mean age, 53.6 years; 47% women; median follow-up, 5.3 years). The final weighted biomarker selection was based on the index prediction accuracy. significantly outperforms standard demographic data longevity (IPA = 29.2 vs. 21.7; 10-year AUC 0.880 0.779; p < 0.001). Age- sex-corrected survival hazard ratio highest-vs-lowest risk quartile 8.73 (95% CI,8.14-9.36) model, increased 24.79 after excluding cancer diagnoses within 5 years CT. Muscle plaque burden, visceral fat density contributed most. Here we show personalized phenotypic can be opportunistically-derived, regardless clinical indication, better inform assessment.

Язык: Английский

---

Association of Obesity and Skeletal Muscle with Postoperative Survival in Non–Small Cell Lung Cancer

Radiology, Год журнала: 2025, Номер 314(1)

Опубликована: Янв. 1, 2025

In patients with non–small cell lung cancer, obesity was associated improved overall survival after curative resection, particularly when CT-assessed skeletal muscle mass and radiodensity were preserved.

Язык: Английский

---

Opportunistic osteoporosis screening using chest CT with artificial intelligence

Osteoporosis International, Год журнала: 2022, Номер 33(12), С. 2547 - 2561

Опубликована: Авг. 6, 2022

Язык: Английский

---

Improved CT-based Osteoporosis Assessment with a Fully Automated Deep Learning Tool

Radiology Artificial Intelligence, Год журнала: 2022, Номер 4(5)

Опубликована: Авг. 31, 2022

To develop, test, and validate a deep learning (DL) tool that improves upon previous feature-based CT image processing bone mineral density (BMD) algorithm compare it against the manual reference standard.This single-center, retrospective, Health Insurance Portability Accountability Act-compliant study included L1 trabecular Hounsfield unit measurements from abdominal scans in 11 035 patients (mean age, 58 years ± 12 [SD]; 6311 women) as standard. Automated level selection region of interest (ROI) placement were then performed this cohort with both previously validated new DL tool. Overall technical success rates agreement standard assessed.The overall rate heterogeneous patient was significantly higher than older BMD (99.3% vs 89.4%, P < .001). Using tool, closest median values for single-, three-, seven-slice vertebral ROIs within 5% 35.1%, 56.9%, 85.8% scans; 10% 56.6%, 75.6%, 92.9% 25% 76.5%, 89.3%, 97.1% scans, respectively. Trade-offs sensitivity specificity osteoporosis assessment observed single-slice approach (sensitivity, 39.4%; specificity, 98.3%) to minimum value multislice (for seven contiguous slices; sensitivity, 71.3% 94.6%).The demonstrated its outputs can be targeted or assessment.Keywords: CT, CT-Quantitative, Abdomen/GI, Skeletal-Axial, Spine, Deep Learning, Machine Learning Supplemental material is available article. © RSNA, 2022.

Язык: Английский

---

AI-based opportunistic CT screening of incidental cardiovascular disease, osteoporosis, and sarcopenia: cost-effectiveness analysis

Abdominal Radiology, Год журнала: 2023, Номер unknown

Опубликована: Янв. 20, 2023

Язык: Английский

---

Multimodality Imaging in Metabolic Syndrome: State-of-the-Art Review

Radiographics, Год журнала: 2024, Номер 44(3)

Опубликована: Фев. 8, 2024

Metabolic syndrome comprises a set of risk factors that include abdominal obesity, impaired glucose tolerance, hypertriglyceridemia, low high-density lipoprotein levels, and high blood pressure, at least three which must be fulfilled for diagnosis. has been linked to an increased cardiovascular disease type 2 diabetes mellitus. Multimodality imaging plays important role in metabolic syndrome, including diagnosis, stratification, assessment complications. CT MRI are the primary tools quantification excess fat, subcutaneous visceral adipose tissue, as well fat around organs, associated with risk. PET shown detect signs insulin resistance may ectopic sites brown fat. Cardiovascular is complication resulting subclinical or symptomatic coronary artery disease, alterations cardiac structure function potential progression heart failure, systemic vascular disease. angiography provides comprehensive evaluation arteries, while assesses structure, function, myocardial ischemia, infarction. Liver damage results from spectrum nonalcoholic fatty liver ranging steatosis fibrosis possible cirrhosis. US, CT, useful assessing can performed grade hepatic fibrosis, particularly using elastography techniques. also deleterious effects on pancreas, kidney, gastrointestinal tract, ovaries, several malignancies. cerebral infarcts, best evaluated MRI, cognitive decline.

Язык: Английский

---