Massimo Falchini

Neural networks for computer-aided diagnosis: detection of lung nodules in chest radiograms

The paper describes a neural-network-based system for the computer aided detection of lung nodules in chest radiograms. Our approach is based on multiscale processing and artificial neural networks (ANNs). The problem of nodule detection is faced by using a two-stage architecture including: 1) an attention focusing subsystem that processes whole radiographs to locate possible nodular regions ensuring high sensitivity; 2) a validation subsystem that processes regions of interest to evaluate the likelihood of the presence of a nodule, so as to reduce false alarms and increase detection specificity. Biologically inspired filters (both LoG and Gabor kernels) are used to enhance salient image features. ANNs of the feedforward type are employed, which allow an efficient use of a priori knowledge about the shape of nodules, and the background structure. The images from the public JSRT database, including 247 radiograms, were used to build and test the system. We performed a further test by using a second private database with 65 radiograms collected and annotated at the Radiology Department of the University of Florence. Both data sets include nodule and nonnodule radiographs. The use of a public data set along with independent testing with a different image set makes the comparison with other systems easier and allows a deeper understanding of system behavior. Experimental results are described by ROC/FROC analysis. For the JSRT database, we observed that by varying sensitivity from 60 to 75% the number of false alarms per image lies in the range 4-10, while accuracy is in the range 95.7-98.0%. When the second data set was used comparable results were obtained. The observed system performances support the undertaking of system validation in clinical settings.

3-D Segmentation Algorithm of Small Lung Nodules in Spiral CT Images

Computed tomography (CT) is the most sensitive imaging technique for detecting lung nodules, and is now being evaluated as a screening tool for lung cancer in several large samples studies all over the world. In this report, we describe a semiautomatic method for 3-D segmentation of lung nodules in CT images for subsequent volume assessment. The distinguishing features of our algorithm are the following. 1) The user interaction process. It allows the introduction of the knowledge of the expert in a simple and reproducible manner. 2) The adoption of the geodesic distance in a multithreshold image representation. It allows the definition of a fusion--segregation process based on both gray-level similarity and objects shape. The algorithm was validated on low-dose CT scans of small nodule phantoms (mean diameter 5.3-11 mm) and in vivo lung nodules (mean diameter 5--9.8 mm) detected in the Italung-CT screening program for lung cancer. A further test on small lung nodules of Lung Image Database Consortium (LIDC) first data set was also performed. We observed a RMS error less than 6.6% in phantoms, and the correct outlining of the nodule contour was obtained in 82/95 lung nodules of Italung-CT and in 10/12 lung nodules of LIDC first data set. The achieved results support the use of the proposed algorithm for volume measurements of lung nodules examined with low-dose CT scanning technique.

Four-year results of low-dose CT screening and nodule management in the ITALUNG trial.

Introduction: Recruitment and nodule management are critical issues of lung cancer screening with low-dose computed tomography (LDCT). We report subjects’ compliance and results of LDCT screening and management protocol in the active arm of the ITALUNG trial. Methods: Three thousand two hundred six smokers or former smokers invited by mail were randomized to receive four annual LDCT (n = 1613) or usual care (n = 1593). Management protocol included follow-up LDCT, 2-[18F]fluoro-2-deoxy-D glucose positron emission tomography (FDG-PET), and CT-guided fine-needle aspiration biopsy (FNAB). Results: One thousand four hundred six subjects (87%) underwent baseline LDCT, and 1263 (79%) completed four screening rounds. LDCT was positive in 30.3% of the subjects at baseline and 15.8% subsequently. Twenty-one lung tumors in 20 subjects (1.5% detection) were found at baseline, and 20 lung tumors in 18 subjects (0.5% detection) in subsequent screening rounds. Ten of 18 prevalent (55%) and 13 of 17 incident (76%) non–small-cell cancers were in stage I. Interval growth enabled diagnosis of lung cancer in 16 subjects (42%), but at least one follow-up LDCT was obtained in 741 subjects (52.7%) over the screening period. FDG-PET obtained in 6.5% of subjects had 84% sensitivity and 90% specificity for malignant lesions. FNAB obtained in 2.4% of subjects showed 90% sensitivity and 88% specificity. Positivity of both FDG-PET and FNAB invariably predicted malignancy. Surgery for benign lesions was performed on four subjects (10% of procedures) but followed protocol violations on three subjects. Conclusions: High-risk subjects recruited by mail who entered LDCT screening showed a high and stable compliance. Efficacy of screening is, however, weakened by low detection rate and specificity. Adhesion to management protocol might lessen surgery for benign lesions.

Risk-benefit analysis of X-ray exposure associated with lung cancer screening in the Italung-CT trial.

OBJECTIVE Prior analyses of X-ray exposures in lung cancer screening with CT considered the basic acquisition technique in single-detector scanners and the effects of a lifetime screening regimen, whereas the potential benefit in terms of lives saved was not addressed. MATERIALS AND METHODS We determined the total-body effective dose of different acquisition techniques for one single-detector and one MDCT scanner and made projections about the cumulative radiation exposure to smokers undergoing four annual CT examinations on the same scanners in the Italung-CT Trial. Combining these data with estimates of radiation-induced fatal cancer and of the benefit of screening, we calculated the risk-benefit ratio for participants in the trial, ex-smokers, and never-smokers. RESULTS The cumulative effective doses per 1,000 subjects were 3.3 Sv using an MDCT scanner and 5.8 or 7.1 Sv using a single-detector scanner. Potential fatal cancers associated with radiation exposure were 0.11 per 1,000 subjects for MDCT scanners and 0.20 or 0.24 for single-detector scanners, which is about 10-100 times lower than the number of expected lives saved by screening assuming a 20-30% lung cancer-specific mortality reduction in current smokers. They were, however, of similar magnitude to the lives saved by screening in never-smokers and former smokers assuming a 10% efficacy of screening. CONCLUSION MDCT is associated with lower radiation doses than single-detector CT technology. The risk of radiation dose in the Italung-CT Trial is compensated for by the expected benefit. CT screening for lung cancer should not be offered to never-smokers, whereas its recommendation in former smokers is debatable.

Automated Segmentation Refinement of Small Lung Nodules in CT Scans by Local Shape Analysis

One of the most important problems in the segmentation of lung nodules in CT imaging arises from possible attachments occurring between nodules and other lung structures, such as vessels or pleura. In this report, we address the problem of vessels attachments by proposing an automated correction method applied to an initial rough segmentation of the lung nodule. The method is based on a local shape analysis of the initial segmentation making use of 3-D geodesic distance map representations. The correction method has the advantage that it locally refines the nodule segmentation along recognized vessel attachments only, without modifying the nodule boundary elsewhere. The method was tested using a simple initial rough segmentation, obtained by a fixed image thresholding. The validation of the complete segmentation algorithm was carried out on small lung nodules, identified in the ITALUNG screening trial and on small nodules of the lung image database consortium (LIDC) dataset. In fully automated mode, 217/256 (84.8%) lung nodules of ITALUNG and 139/157 (88.5%) individual marks of lung nodules of LIDC were correctly outlined and an excellent reproducibility was also observed. By using an additional interactive mode, based on a controlled manual interaction, 233/256 (91.0%) lung nodules of ITALUNG and 144/157 (91.7%) individual marks of lung nodules of LIDC were overall correctly segmented. The proposed correction method could also be usefully applied to any existent nodule segmentation algorithm for improving the segmentation quality of juxta-vascular nodules.

Dose exposure in the ITALUNG trial of lung cancer screening with low-dose CT

Few data are available on the effective dose received by participants in lung cancer screening programmes with low-dose CT (LDCT). We report the collective effective dose delivered to 1406 current or former smokers enrolled in the ITALUNG trial who completed 4 annual LDCT examinations and related further investigations including follow-up LDCT, 2-[(18)F]flu-2-deoxy-d-glucose positron emission tomography (FDG-PET) or CT-guided fine needle aspiration biopsy (FNAB). Using the air CT dose index and Monte Carlo simulations on an anthropomorphic phantom, the whole-body effective dose associated with LDCT was determined for the eight CT scanners used in the trial. A value of 7 mSv was assigned to FDG-PET while the measured mean effective dose of CT-guided FNAB was 1.5 mSv. The mean collective effective dose in the 1406 subjects ranged between 8.75 and 9.36 Sv and the mean effective dose to the single subject over 4 years was between 6.2 and 6.8 mSv (range 1.7-21.5 mSv) according to the cranial-caudal length of the LDCT volume. 77.4% of the dose was owing to annual LDCT and 22.6% to further investigations. Considering the nominal risk coefficients for stochastic effects after exposure to low-dose radiation according to the National Radiological Protection Board, International Commission on Radiological Protection (ICRP) 60, ICRP103 and Biological Effects of Ionizing Radiation VII, the mean number of radiation-induced cancers ranged between 0.12 and 0.33 per 1000 subjects. The individual effective dose to participants in a 4-year lung cancer screening programme with annual LDCT is very low and about one-third of the effective dose that is associated with natural background radiation and diagnostic radiology in the same time period.

The

Nodule growth as observed in computed tomography (CT) scans acquired at different times is the primary feature to malignancy of indeterminate small lung nodules. In this paper, we propose the estimation of nodule size through a scale-space representation which needs no segmentation and has high intra- and inter-operator reproducibility. Lung nodules usually appear in CT images as blob-like patterns and can be analyzed in the scale-space by Laplacian of Gaussian (LoG ) kernels. For each nodular pattern the LoG scale-space signature was computed and the related characteristic scale adopted as measurement of nodule size. Both in vitro and in vivo validation of LoG characteristic scale were carried out. In vitro validation was done by 40 nondeformable phantoms and 10 deformable phantoms. A close relationship between the characteristic scale and the equivalent diameter, i.e., the diameter of the sphere having the same volume of nodules, (Pearson correlation coefficient was 0.99) and, for nodules undergoing little deformations (obtained at constant volume), small variability of the characteristic scale was observed. The in vivo validation was performed on low and standard-dose CT scans collected from the ITALUNG screening trial (86 nodules) and from the LIDC public data set (89 solid nodules and 40 part-solid nodules or ground-glass opacities). The Pearson correlation coefficient between characteristic scale and equivalent diameter was 0.83-0.93 for ITALUNG and 0.68-0.83 for LIDC data set. Intra- and inter-operator reproducibility of characteristic scale was excellent: on a set of 40 lung nodules of ITALUNG data, two radiologists produced identical results in repeated measurements. The scan-rescan variability of the characteristic scale was also investigated on 86 two-year-stable solid lung nodules (each one observed, on average, in four CT scans) identified in the ITALUNG screening trial: a coefficient of repeatability of about 0.9 mm was observed. Experimental evidence supports the clinical use of the LoG characteristic scale to measure nodule size in CT imaging.

{LoG}

BACKGROUND Population screening for colorectal cancer (CRC) is widely adopted, but the preferred strategy is still under debate. We aimed to compare reduced (r-CTC) and full cathartic preparation CT colonography (f-CTC), fecal immunochemical test (FIT), and optical colonoscopy (OC) as primary screening tests for CRC. METHODS Citizens of a district of Florence, Italy, age 54 to 65 years, were allocated (8:2.5:2.5:1) with simple randomization to be invited by mail to one of four screening interventions: 1) biennial FIT for three rounds, 2) r-CTC, 3) f-CTC, 4) OC. Patients tested positive to FIT or CTC (at least one polyp ≥6mm) were referred to OC work-up. The primary outcomes were participation rate and detection rate (DR) for cancer or advanced adenoma (advanced neoplasia). All statistical tests were two-sided. RESULTS Sixteen thousand eighty-seven randomly assigned subjects were invited to the assigned screening test. Participation rates were 50.4% (4677/9288) for first-round FIT, 28.1% (674/2395) for r-CTC, 25.2% (612/2430) for f-CTC, and 14.8% (153/1036) for OC. All differences between groups were statistically significant (P = .047 for r-CTC vs f-CTC; P < .001 for all others). DRs for advanced neoplasia were 1.7% (79/4677) for first-round FIT, 5.5% (37/674) for r-CTC, 4.9% (30/612) for f-CTC, and 7.2% (11/153) for OC. Differences in DR between CTC groups and FIT were statistically significant (P < .001), but not between r-CTC and f-CTC (P = .65). CONCLUSIONS Reduced preparation increases participation in CTC. Lower attendance and higher DR of CTC as compared with FIT are key factors for the optimization of its role in population screening of CRC.

Characteristic Scale: A Consistent Measurement of Lung Nodule Size in CT Imaging

BackgroundColorectal cancer (CRC) is the most frequent cancer in Europe. Randomized clinical trials demonstrated that screening with fecal occult blood test (FOBT) reduces mortality from CRC. Accordingly, the European Community currently recommends population-based screening with FOBT. Other screening tests, such as computed tomography colonography (CTC) and optical colonoscopy (OC), are highly accurate for examining the entire colon for adenomas and CRC. Acceptability represents a critical determinant of the impact of a screening program. We designed a randomized controlled trial to compare participation rate and diagnostic yield of FOBT, CTC with computer-aided diagnosis, and OC as primary tests for population-based screening.Methods/DesignA total of 14,000 subjects aged 55 to 64 years, living in the Florence district and never screened for CRC, will be randomized in three arms: group 1 (5,000 persons) invited to undergo CTC (divided into: subgroup 1A with reduced cathartic preparation and subgroup 1B with standard bowel preparation); group 2 (8,000 persons) invited to undergo a biannual FOBT for three rounds; and group 3 (1,000 persons) invited to undergo OC. Subjects of each group will be invited by mail to undergo the selected test. All subjects with a positive FOBT or CTC test (that is, mass or at least one polyp ≥6 mm) will be invited to undergo a second-level OC. Primary objectives of the study are to compare the participation rate to FOBT, CTC and OC; to compare the detection rate for cancer or advanced adenomas of CTC versus three rounds of biannual FOBT; to evaluate referral rate for OC induced by primary CTC versus three rounds of FOBT; and to estimate costs of the three screening strategies. A secondary objective of the study is to create a biological bank of blood and stool specimens from subjects undergoing CTC and OC.DiscussionThis study will provide information about participation/acceptability, diagnostic yield and costs of screening with CTC in comparison with the recommended test (FOBT) and OC.Trial registrationClinicalTrials.gov Identifier: NCT01651624.

Reduced and Full-Preparation CT Colonography, Fecal Immunochemical Test, and Colonoscopy for Population Screening of Colorectal Cancer: A Randomized Trial

BACKGROUND Refusal of colonoscopy is a drawback of colorectal cancer screening programmes based on faecal occult blood test. Computed-tomographic-colonography is generally more accepted than colonoscopy. AIM To compare adherence to computed-tomographic-colonography and second-invitation colonoscopy in subjects with positive faecal test refusing colonoscopy. METHODS We performed a prospective study in 198 subjects with positive faecal test who refused first referral to colonoscopy in one endoscopy service of the Florence screening programme. Subjects were randomly invited to computed-tomographic-colonography (n = 100) or re-invited to colonoscopy (n = 98). Mail invitation was followed by a questionnaire administered by phone. Computed-tomographic-colonography findings were verified with colonoscopy. RESULTS 32 subjects could not be reached, 71 (35.9%) had undergone colonoscopy on their own; 4 were excluded for contraindications; 30/48 (62.5%) in the computed-tomographic-colonography arm and 11/43 (25.6%) in the colonoscopy arm accepted the proposed examinations (p < 0.001). Four advanced adenomas and 1 cancer were found in the 28 subjects who ultimately underwent computed-tomographic-colonography and 2 advanced adenomas and 2 cancers in the 9 subjects who ultimately underwent second-invitation colonoscopy. CONCLUSION Subjects with positive faecal occult blood test refusing colonoscopy show a higher adherence to computed-tomographic-colonography than to second invitation colonoscopy.