RESUMO
OBJECTIVES: To evaluate and compare the measurement accuracy of two different computer-aided diagnosis (CAD) systems regarding artificial pulmonary nodules and assess the clinical impact of volumetric inaccuracies in a phantom study. METHODS: In this phantom study, 59 different phantom arrangements with 326 artificial nodules (178 solid, 148 ground-glass) were scanned at 80 kV, 100 kV, and 120 kV. Four different nodule diameters were used: 5 mm, 8 mm, 10 mm, and 12 mm. Scans were analyzed by a deep-learning (DL)-based CAD and a standard CAD system. Relative volumetric errors (RVE) of each system vs. ground truth and the relative volume difference (RVD) DL-based vs. standard CAD were calculated. The Bland-Altman method was used to define the limits of agreement (LOA). The hypothetical impact on LungRADS classification was assessed for both systems. RESULTS: There was no difference between the three voltage groups regarding nodule volumetry. Regarding the solid nodules, the RVE of the 5-mm-, 8-mm-, 10-mm-, and 12-mm-size groups for the DL CAD/standard CAD were 12.2/2.8%, 1.3/ - 2.8%, - 3.6/1.5%, and - 12.2/ - 0.3%, respectively. The corresponding values for the ground-glass nodules (GGN) were 25.6%/81.0%, 9.0%/28.0%, 7.6/20.6%, and 6.8/21.2%. The mean RVD for solid nodules/GGN was 1.3/ - 15.2%. Regarding the LungRADS classification, 88.5% and 79.8% of all solid nodules were correctly assigned by the DL CAD and the standard CAD, respectively. 14.9% of the nodules were assigned differently between the systems. CONCLUSIONS: Patient management may be affected by the volumetric inaccuracy of the CAD systems and hence demands supervision and/or manual correction by a radiologist. KEY POINTS: ⢠The DL-based CAD system was more accurate in the volumetry of GGN and less accurate regarding solid nodules than the standard CAD system. ⢠Nodule size and attenuation have an effect on the measurement accuracy of both systems; tube voltage has no effect on measurement accuracy. ⢠Measurement inaccuracies of CAD systems can have an impact on patient management, which demands supervision by radiologists.
Assuntos
Neoplasias Pulmonares , Nódulos Pulmonares Múltiplos , Nódulo Pulmonar Solitário , Humanos , Tomografia Computadorizada por Raios X/métodos , Diagnóstico por Computador/métodos , Nódulos Pulmonares Múltiplos/diagnóstico por imagem , Imagens de Fantasmas , Radiologistas , Neoplasias Pulmonares/diagnóstico por imagem , Nódulo Pulmonar Solitário/diagnóstico por imagem , Nódulo Pulmonar Solitário/terapia , Interpretação de Imagem Radiográfica Assistida por Computador/métodos , Sensibilidade e EspecificidadeRESUMO
OBJECTIVES: To assess the value of quantitative computed tomography (QCT) of the whole lung and nodule-bearing lobe regarding pulmonary nodule malignancy risk estimation. METHODS: A total of 251 subjects (median [IQR] age, 65 (57-73) years; 37% females) with pulmonary nodules on non-enhanced thin-section CT were retrospectively included. Twenty percent of the nodules were malignant, the remainder benign either histologically or at least 1-year follow-up. CT scans were subjected to in-house software, computing parameters such as mean lung density (MLD) or peripheral emphysema index (pEI). QCT variable selection was performed using logistic regression; selected variables were integrated into the Mayo Clinic and the parsimonious Brock Model. RESULTS: Whole-lung analysis revealed differences between benign vs. malignant nodule groups in several parameters, e.g. the MLD (-766 vs. -790 HU) or the pEI (40.1 vs. 44.7 %). The proposed QCT model had an area-under-the-curve (AUC) of 0.69 (95%-CI, 0.62-0.76) based on all available data. After integrating MLD and pEI into the Mayo Clinic and Brock Model, the AUC of both clinical models improved (AUC, 0.91 to 0.93 and 0.88 to 0.91, respectively). The lobe-specific analysis revealed that the nodule-bearing lobes had less emphysema than the rest of the lung regarding benign (EI, 0.5 vs. 0.7 %; p < 0.001) and malignant nodules (EI, 1.2 vs. 1.7 %; p = 0.001). CONCLUSIONS: Nodules in subjects with higher whole-lung metrics of emphysema and less fibrosis are more likely to be malignant; hereby the nodule-bearing lobes have less emphysema. QCT variables could improve the risk assessment of incidental pulmonary nodules. KEY POINTS: ⢠Nodules in subjects with higher whole-lung metrics of emphysema and less fibrosis are more likely to be malignant. ⢠The nodule-bearing lobes have less emphysema compared to the rest of the lung. ⢠QCT variables could improve the risk assessment of incidental pulmonary nodules.
Assuntos
Enfisema , Neoplasias Pulmonares , Nódulos Pulmonares Múltiplos , Enfisema Pulmonar , Nódulo Pulmonar Solitário , Feminino , Humanos , Idoso , Masculino , Estudos Retrospectivos , Nódulo Pulmonar Solitário/diagnóstico por imagem , Nódulo Pulmonar Solitário/patologia , Pulmão/diagnóstico por imagem , Pulmão/patologia , Nódulos Pulmonares Múltiplos/diagnóstico por imagem , Nódulos Pulmonares Múltiplos/patologia , Enfisema Pulmonar/diagnóstico por imagem , Enfisema Pulmonar/patologia , Tomografia Computadorizada por Raios X/métodos , Neoplasias Pulmonares/diagnóstico por imagem , Neoplasias Pulmonares/patologia , FibroseRESUMO
OBJECTIVES: The purpose of this study was to determine the influence of dose reduction on a commercially available lung cancer prediction convolutional neuronal network (LCP-CNN). METHODS: CT scans from a cohort provided by the local lung cancer center (n = 218) with confirmed pulmonary malignancies and their corresponding reduced dose simulations (25% and 5% dose) were subjected to the LCP-CNN. The resulting LCP scores (scale 1-10, increasing malignancy risk) and the proportion of correctly classified nodules were compared. The cohort was divided into a low-, medium-, and high-risk group based on the respective LCP scores; shifts between the groups were studied to evaluate the potential impact on nodule management. Two different malignancy risk score thresholds were analyzed: a higher threshold of ≥ 9 ("rule-in" approach) and a lower threshold of > 4 ("rule-out" approach). RESULTS: In total, 169 patients with 196 nodules could be included (mean age ± SD, 64.5 ± 9.2 year; 49% females). Mean LCP scores for original, 25% and 5% dose levels were 8.5 ± 1.7, 8.4 ± 1.7 (p > 0.05 vs. original dose) and 8.2 ± 1.9 (p < 0.05 vs. original dose), respectively. The proportion of correctly classified nodules with the "rule-in" approach decreased with simulated dose reduction from 58.2 to 56.1% (p = 0.34) and to 52.0% for the respective dose levels (p = 0.01). For the "rule-out" approach the respective values were 95.9%, 96.4%, and 94.4% (p = 0.12). When reducing the original dose to 25%/5%, eight/twenty-two nodules shifted to a lower, five/seven nodules to a higher malignancy risk group. CONCLUSION: CT dose reduction may affect the analyzed LCP-CNN regarding the classification of pulmonary malignancies and potentially alter pulmonary nodule management. CLINICAL RELEVANCE STATEMENT: Utilization of a "rule-out" approach with a lower malignancy risk threshold prevents underestimation of the nodule malignancy risk for the analyzed software, especially in high-risk cohorts. KEY POINTS: ⢠LCP-CNN may be affected by CT image parameters such as noise resulting from low-dose CT acquisitions. ⢠CT dose reduction can alter pulmonary nodule management recommendations by affecting the outcome of the LCP-CNN. ⢠Utilization of a lower malignancy risk threshold prevents underestimation of pulmonary malignancies in high-risk cohorts.
RESUMO
OBJECTIVES: This study was conducted to evaluate the effect of dose reduction on the performance of a deep learning (DL)-based computer-aided diagnosis (CAD) system regarding pulmonary nodule detection in a virtual screening scenario. METHODS: Sixty-eight anthropomorphic chest phantoms were equipped with 329 nodules (150 ground glass, 179 solid) with four sizes (5 mm, 8 mm, 10 mm, 12 mm) and scanned with nine tube voltage/current combinations. The examinations were analyzed by a commercially available DL-based CAD system. The results were compared by a comparison of proportions. Logistic regression was performed to evaluate the impact of tube voltage, tube current, nodule size, nodule density, and nodule location. RESULTS: The combination with the lowest effective dose (E) and unimpaired detection rate was 80 kV/50 mAs (sensitivity: 97.9%, mean false-positive rate (FPR): 1.9, mean CTDIvol: 1.2 ± 0.4 mGy, mean E: 0.66 mSv). Logistic regression revealed that tube voltage and current had the greatest impact on the detection rate, while nodule size and density had no significant influence. CONCLUSIONS: The optimal tube voltage/current combination proposed in this study (80 kV/50 mAs) is comparable to the proposed combinations in similar studies, which mostly dealt with conventional CAD software. Modification of tube voltage and tube current has a significant impact on the performance of DL-based CAD software in pulmonary nodule detection regardless of their size and composition. KEY POINTS: ⢠Modification of tube voltage and tube current has a significant impact on the performance of deep learning-based CAD software. ⢠Nodule size and composition have no significant impact on the software's performance. ⢠The optimal tube voltage/current combination for the examined software is 80 kV/50 mAs.
Assuntos
Aprendizado Profundo , Neoplasias Pulmonares , Nódulo Pulmonar Solitário , Algoritmos , Humanos , Neoplasias Pulmonares/diagnóstico por imagem , Doses de Radiação , Interpretação de Imagem Radiográfica Assistida por Computador/métodos , Sensibilidade e Especificidade , Nódulo Pulmonar Solitário/diagnóstico por imagem , Tomografia Computadorizada por Raios X/métodosRESUMO
Background and Objectives: Osteoarthritis (OA) of the knee is a degenerative disorder characterized by damage to the joint cartilage, pain, swelling, and walking disability. The purpose of this study was to assess whether demographic and radiologic parameters (knee diameters and knee cross-sectional area from magnetic resonance (MR) images) could be used as surrogate biomarkers for the prediction of OA. Materials and Methods: The knee diameters and cross-sectional areas of 481 patients were measured on knee MR images, and the corresponding demographic parameters were extracted from the patients' clinical records. The images were graded based on the modified Outerbridge arthroscopic classification that was used as ground truth. Receiver-operating characteristic (ROC) analysis was performed on the collected data. Results: ROC analysis established that age was the most accurate predictor of severe knee cartilage degeneration (corresponding to Outerbridge grades 3 and 4) with an area under the curve (AUC) of the specificity-sensitivity plot of 0.865 ± 0.02. An age over 41 years was associated with a sensitivity and specificity for severe degeneration of 82.8% (CI: 77.5-87.3%), and 76.4% (CI: 70.4-81.6%), respectively. The second-best degeneration predictor was the normalized knee cross-sectional area, with an AUC of 0.767 ± 0.04), followed by BMI (AUC = 0.739 ± 0.02), and normalized knee maximal diameter (AUC = 0.724 ± 0.05), meaning that knee degeneration increases with increasing knee diameter. Conclusions: Age is the best predictor of knee damage progression in OA and can be used as surrogate marker for knee degeneration. Knee diameters and cross-sectional area also correlate with the extent of cartilage lesions. Though less-accurate predictors of damage progression than age, they have predictive value and are therefore easily available surrogate markers of OA that can be used also by general practitioners and orthopedic surgeons.
Assuntos
Doenças das Cartilagens , Cartilagem Articular , Osteoartrite do Joelho , Humanos , Adulto , Osteoartrite do Joelho/diagnóstico por imagem , Articulação do Joelho/diagnóstico por imagem , Articulação do Joelho/cirurgia , Imageamento por Ressonância Magnética/métodos , Biomarcadores , Doenças das Cartilagens/diagnóstico por imagem , Doenças das Cartilagens/patologia , Cartilagem Articular/diagnóstico por imagem , Cartilagem Articular/patologiaRESUMO
OBJECTIVES: To obtain an overview of the attitudes toward interdisciplinary further education of residents and consultants in radiology and nuclear medicine and preferences regarding a future joint training curriculum in Switzerland. METHODS: A 34-item questionnaire was sent electronically (SurveyMonkey online survey tool) to 1244 radiologists and nuclear physicians (residents and consultants) in Switzerland. The items asked about the motivation for further education in each other's specialty and preferences regarding a joint further education curriculum in radiology and nuclear medicine. RESULTS: Overall, 370 questionnaires were analyzed (370/1244, 30%). There were 280 (76%) board-certified physicians in either radiology (238/370, 64%) or nuclear medicine (42/370, 12%) and 65 (18%) residents (radiology 54/370, 15%; nuclear medicine 11/370, 3%). More than half of all residents (34/65, 52%) stated their conviction that a wide range of expertise in both disciplines could be fully guaranteed through adequate cross-curricular training. For responders already at a consultant level in radiology or nuclear medicine, the willingness to undergo further training in each other's specialty significantly increased with a shorter training period. The preferred option for a possible future joint training curriculum was a combination of a 5-year radiology training program with 2 years of further training in nuclear medicine. CONCLUSIONS: Both residents and board-certified physicians in Switzerland are highly interested in a cross-curricular training curriculum in radiology and nuclear medicine. KEY POINTS: ⢠A systematic survey was conducted to obtain information on interest in cross-curricular training in radiology and nuclear medicine and preferences regarding a future joint training curriculum. ⢠More than half of radiology and nuclear medicine residents would be interested in further training in the other specialty. ⢠There is a strong desire for a shorter training program when combining training in both radiology and nuclear medicine.
Assuntos
Currículo , Medicina Nuclear/educação , Radiologia/educação , Adulto , Consultores , Educação Médica Continuada , Feminino , Humanos , Internato e Residência , Masculino , Pessoa de Meia-Idade , Inquéritos e Questionários , Suíça , Adulto JovemRESUMO
ABSTRACT: Interstitial lung disease (ILD) is now diagnosed by an ILD-board consisting of radiologists, pulmonologists, and pathologists. They discuss the combination of computed tomography (CT) images, pulmonary function tests, demographic information, and histology and then agree on one of the 200 ILD diagnoses. Recent approaches employ computer-aided diagnostic tools to improve detection of disease, monitoring, and accurate prognostication. Methods based on artificial intelligence (AI) may be used in computational medicine, especially in image-based specialties such as radiology. This review summarises and highlights the strengths and weaknesses of the latest and most significant published methods that could lead to a holistic system for ILD diagnosis. We explore current AI methods and the data use to predict the prognosis and progression of ILDs. It is then essential to highlight the data that holds the most information related to risk factors for progression, e.g., CT scans and pulmonary function tests. This review aims to identify potential gaps, highlight areas that require further research, and identify the methods that could be combined to yield more promising results in future studies.
Assuntos
Inteligência Artificial , Doenças Pulmonares Intersticiais , Humanos , Doenças Pulmonares Intersticiais/diagnóstico por imagem , Prognóstico , Tomografia Computadorizada por Raios X/métodos , Radiologistas , Pulmão/diagnóstico por imagemRESUMO
The aim of this study was to investigate the diagnostic accuracy and reader confidence for late-gadolinium enhancement (LGE) detection of a novel free-breathing, image-based navigated 3D whole-heart LGE sequence with fat-water separation, compared to a free-breathing motion-corrected 2D LGE sequence in patients with ischemic and non-ischemic cardiomyopathy. Cardiac MRI patients including the respective sequences were retrospectively included. Two independent, blinded readers rated image quality, depiction of segmental LGE and documented acquisition time, SNR, CNR and amount of LGE. Results were compared using the Friedman or the Kruskal-Wallis test. For LGE rating, a jackknife free-response receiver operating characteristic analysis was performed with a figure of merit (FOM) calculation. Forty-two patients were included, thirty-two were examined with a 1.5 T-scanner and ten patients with a 3 T-scanner. The mean acquisition time of the 2D sequence was significantly shorter compared to the 3D sequence (07:12 min vs. 09:24 min; p < 0.001). The 3D scan time was significantly shorter when performed at 3 T compared to 1.5 T (07:47 min vs. 09:50 min; p < 0.001). There were no differences regarding SNR, CNR or amount of LGE. 3D imaging had a significantly higher FOM (0.89 vs. 0.78; p < 0.001). Overall image quality ratings were similar, but 3D sequence ratings were higher for fine anatomical structures. Free-breathing motion-corrected 3D LGE with high isotropic resolution results in enhanced LGE-detection with higher confidence and better delineation of fine structures. The acquisition time for 3D imaging was longer, but may be reduced by performing on a 3 T-scanner.
Assuntos
Meios de Contraste , Gadolínio , Humanos , Cicatriz , Água , Estudos Retrospectivos , Valor Preditivo dos Testes , Imageamento por Ressonância Magnética/métodos , Imageamento Tridimensional/métodos , Aumento da Imagem/métodosRESUMO
Despite current recommendations, there is no recent scientific study comparing the influence of CT reconstruction kernels on lung pattern recognition in interstitial lung disease (ILD).To evaluate the sensitivity of lung (i70) and soft (i30) CT kernel algorithms for the diagnosis of ILD patterns.We retrospectively extracted between 15-25 pattern annotations per case (1 annotationâ=â15 slices of 1âmm) from 23 subjects resulting in 408 annotation stacks per lung kernel and soft kernel reconstructions. Two subspecialized chest radiologists defined the ground truth in consensus. 4 residents, 2 fellows, and 2 general consultants in radiology with 3 to 13 years of experience in chest imaging performed a blinded readout. In order to account for data clustering, a generalized linear mixed model (GLMM) with random intercept for reader and nested for patient and image and a kernel/experience interaction term was used to analyze the results.The results of the GLMM indicated, that the odds of correct pattern recognition is 12â% lower with lung kernel compared to soft kernel; however, this was not statistically significant (OR 0.88; 95%-CI, 0.73-1.06; pâ=â0.187). Furthermore, the consultants' odds of correct pattern recognition was 78â% higher than the residents' odds, although this finding did not reach statistical significance either (OR 1.78; 95%-CI, 0.62-5.06; pâ=â0.283). There was no significant interaction between the two fixed terms kernel and experience. Intra-rater agreement between lung and soft kernel was substantial (κâ=â0.63â±â0.19). The mean inter-rater agreement for lung/soft kernel was κâ=â0.37â±â0.17/κâ=â0.38â±â0.17.There is no significant difference between lung and soft kernel reconstructed CT images for the correct pattern recognition in ILD. There are non-significant trends indicating that the use of soft kernels and a higher level of experience lead to a higher probability of correct pattern identification. · There is no significant difference between lung and soft kernel reconstructed CT images for the correct pattern recognition in interstitial lung disease.. · There are even non-significant tendencies that the use of soft kernels lead to a higher probability of correct pattern identification.. · These results challenge the current recommendations and the routinely performed separate lung kernel reconstructions for lung parenchyma analysis.. CITATION FORMAT: · Klaus JB, Christodoulidis S, Peters AA etâal. Influence of Lung Reconstruction Algorithms on Interstitial Lung Pattern Recognition on CT. Fortschr Röntgenstr 2023; 195: 47â-â54.
Assuntos
Pulmão , Tomografia Computadorizada por Raios X , Humanos , Tomografia Computadorizada por Raios X/métodos , Estudos Retrospectivos , Pulmão/diagnóstico por imagem , Intensificação de Imagem Radiográfica/métodos , AlgoritmosRESUMO
Following intravenous contrast medium (CM) injection, a small proportion of patients acquires hypersensitivity reactions that occur either immediately or non-immediately (delayed). Although it is now claer that even oral applied CMs are able to cause adverse reactions, many radiologists as well as physicians of other disciplines, still believe that CM-application via the gastrointestinal route does not induce hypersensitivity reactions. Since this kind of misinterpretation may harm the patient, education on this topic is still necessary. Therefore, we describe a case who acquired a delayed hypersensitivity reaction following the oral intake of a non-ionic iodinated CM.
Assuntos
Hipersensibilidade a Drogas , Hipersensibilidade Tardia , Hipersensibilidade Imediata , Meios de Contraste/efeitos adversos , Hipersensibilidade a Drogas/diagnóstico , Hipersensibilidade a Drogas/etiologia , Humanos , Hipersensibilidade Tardia/induzido quimicamente , Hipersensibilidade Tardia/complicações , Hipersensibilidade Tardia/diagnóstico , Hipersensibilidade Imediata/induzido quimicamente , Hipersensibilidade Imediata/complicações , Hipersensibilidade Imediata/diagnósticoRESUMO
BACKGROUND: Despite the decreasing relevance of chest radiography in lung cancer screening, chest radiography is still frequently applied to assess for lung nodules. The aim of the current study was to determine the accuracy of a commercial AI based CAD system for the detection of artificial lung nodules on chest radiograph phantoms and compare the performance to radiologists in training. METHODS: Sixty-one anthropomorphic lung phantoms were equipped with 140 randomly deployed artificial lung nodules (5, 8, 10, 12 mm). A random generator chose nodule size and distribution before a two-plane chest X-ray (CXR) of each phantom was performed. Seven blinded radiologists in training (2 fellows, 5 residents) with 2 to 5 years of experience in chest imaging read the CXRs on a PACS-workstation independently. Results of the software were recorded separately. McNemar test was used to compare each radiologist's results to the AI-computer-aided-diagnostic (CAD) software in a per-nodule and a per-phantom approach and Fleiss-Kappa was applied for inter-rater and intra-observer agreements. RESULTS: Five out of seven readers showed a significantly higher accuracy than the AI algorithm. The pooled accuracies of the radiologists in a nodule-based and a phantom-based approach were 0.59 and 0.82 respectively, whereas the AI-CAD showed accuracies of 0.47 and 0.67, respectively. Radiologists' average sensitivity for 10 and 12 mm nodules was 0.80 and dropped to 0.66 for 8 mm (P=0.04) and 0.14 for 5 mm nodules (P<0.001). The radiologists and the algorithm both demonstrated a significant higher sensitivity for peripheral compared to central nodules (0.66 vs. 0.48; P=0.004 and 0.64 vs. 0.094; P=0.025, respectively). Inter-rater agreements were moderate among the radiologists and between radiologists and AI-CAD software (K'=0.58±0.13 and 0.51±0.1). Intra-observer agreement was calculated for two readers and was almost perfect for the phantom-based (K'=0.85±0.05; K'=0.80±0.02); and substantial to almost perfect for the nodule-based approach (K'=0.83±0.02; K'=0.78±0.02). CONCLUSIONS: The AI based CAD system as a primary reader acts inferior to radiologists regarding lung nodule detection in chest phantoms. Chest radiography has reasonable accuracy in lung nodule detection if read by a radiologist alone and may be further optimized by an AI based CAD system as a second reader.
RESUMO
Late post-transplant Pneumocystis jirovecii pneumonia (PcP) has been reported in many renal transplant recipients (RTRs) centers using universal prophylaxis. Specific features of PcP compared to other respiratory infections in the same population are not well reported. We analyzed clinical, laboratory, administrative and radiological data of all confirmed PcP cases between January 2009 and December 2014. To identify factors specifically associated with PcP, we compared clinical and laboratory data of RTRs with non-PcP. Over the study period, 36 cases of PcP were identified. Respiratory distress was more frequent in PcP compared to non-PcP (tachypnea: 59%, 20/34 vs. 25%, 13/53, p = 0.0014; dyspnea: 70%, 23/33 vs. 44%, 24/55, p = 0.0181). In contrast, fever was less frequent in PcP compared to non-PcP pneumonia (35%, 11/31 vs. 76%, 42/55, p = 0.0002). In both cohorts, total lymphocyte count and serum sodium decreased, whereas lactate dehydrogenase (LDH) increased at diagnosis. Serum calcium increased in PcP and decreased in non-PcP. In most PcP cases (58%, 21/36), no formal indication for restart of PcP prophylaxis could be identified. Potential transmission encounters, suggestive of interhuman transmission, were found in 14/36, 39% of patients. Interhuman transmission seems to contribute importantly to PcP among RTRs. Hypercalcemia, but not elevated LDH, was associated with PcP when compared to non-PcP.
RESUMO
OBJECTIVES: The objective of this study is to assess the performance of a computer-aided diagnosis (CAD) system (INTACT system) for the automatic classification of high-resolution computed tomography images into 4 radiological diagnostic categories and to compare this with the performance of radiologists on the same task. MATERIALS AND METHODS: For the comparison, a total of 105 cases of pulmonary fibrosis were studied (54 cases of nonspecific interstitial pneumonia and 51 cases of usual interstitial pneumonia). All diagnoses were interstitial lung disease board consensus diagnoses (radiologically or histologically proven cases) and were retrospectively selected from our database. Two subspecialized chest radiologists made a consensual ground truth radiological diagnosis, according to the Fleischner Society recommendations. A comparison analysis was performed between the INTACT system and 2 other radiologists with different years of experience (readers 1 and 2). The INTACT system consists of a sequential pipeline in which first the anatomical structures of the lung are segmented, then the various types of pathological lung tissue are identified and characterized, and this information is then fed to a random forest classifier able to recommend a radiological diagnosis. RESULTS: Reader 1, reader 2, and INTACT achieved similar accuracy for classifying pulmonary fibrosis into the original 4 categories: 0.6, 0.54, and 0.56, respectively, with P > 0.45. The INTACT system achieved an F-score (harmonic mean for precision and recall) of 0.56, whereas the 2 readers, on average, achieved 0.57 (P = 0.991). For the pooled classification (2 groups, with and without the need for biopsy), reader 1, reader 2, and CAD had similar accuracies of 0.81, 0.70, and 0.81, respectively. The F-score was again similar for the CAD system and the radiologists. The CAD system and the average reader reached F-scores of 0.80 and 0.79 (P = 0.898). CONCLUSIONS: We found that a computer-aided detection algorithm based on machine learning was able to classify idiopathic pulmonary fibrosis with similar accuracy to a human reader.