Does a computer-aided detection algorithm in a second read paradigm enhance the performance of experienced computed tomography colonography readers in a population of increased risk?

Haustral fold segmentation with curvature-guided level set evolution

Human colon has complex structures mostly because of the haustral folds. The folds are thin flat protrusions on the colon wall, which complicate the shape analysis for computer-aided detection (CAD) of colonic polyps. Fold segmentation may help reduce the structural complexity, and the folds can serve as an anatomic reference for computed tomographic colonography (CTC). Therefore, in this study, based on a model of the haustral fold boundaries, we developed a level-set approach to automatically segment the fold surfaces. To evaluate the developed fold segmentation algorithm, we first established the ground truth of haustral fold boundaries by experts' drawing on 15 patient CTC datasets without severe under/over colon distention from two medical centers. The segmentation algorithm successfully detected 92.7% of the folds in the ground truth. In addition to the sensitivity measure, we further developed a merit of segmented-area ratio (SAR), i.e., the ratio between the area of the intersection and union of the expert-drawn folds and the area of the automatically segmented folds, to measure the segmentation accuracy. The segmentation algorithm reached an average value of SAR = 86.2%, showing a good match with the ground truth on the fold surfaces. We believe the automatically segmented fold surfaces have the potential to benefit many postprocedures in CTC, such as CAD, taenia coli extraction, supine-prone registration, etc.

The second ESGAR consensus statement on CT colonography

Objective

To update quality standards for CT colonography based on consensus among opinion leaders within the European Society of Gastrointestinal and Abdominal Radiology (ESGAR).

Material and methods

A multinational European panel of nine members of the ESGAR CT colonography Working Group (representing six EU countries) used a modified Delphi process to rate their level of agreement on a variety of statements pertaining to the acquisition, interpretation and implementation of CT colonography. Four Delphi rounds were conducted, each at 2 months interval.

Results

The panel elaborated 86 statements.

In the final round the panelists achieved complete consensus in 71 of 86 statements (82 %). Categories including the highest proportion of statements with excellent Cronbach's internal reliability were colon distension, scan parameters, use of intravenous contrast agents, general guidelines on patient preparation, role of CAD and lesion measurement.

Lower internal reliability was achieved for the use of a rectal tube, spasmolytics, decubitus positioning and number of CT data acquisitions, faecal tagging, 2D vs. 3D reading, and reporting.

Conclusion

The recommendations of the consensus should be useful for both the radiologist who is starting a CTC service and for those who have already implemented the technique but whose practice may need updating.

Key Points

• Computed tomographic colonography is the optimal radiological method of assessing the colon

• This article reviews ESGAR quality standards for CT colonography

• This article is aimed to provide CT-colonography guidelines for practising radiologists

• The recommendations should help radiologists who are starting/updating their CTC services

ACRIN CT colonography trial: does reader's preference for primary two-dimensional versus primary three-dimensional interpretation affect performance?

PURPOSE

To determine whether the reader's preference for a primary two-dimensional (2D) or three-dimensional (3D) computed tomographic (CT) colonographic interpretation method affects performance when using each technique.

MATERIALS AND METHODS

In this institutional review board-approved, HIPAA-compliant study, images from 2531 CT colonographic examinations were interpreted by 15 trained radiologists by using colonoscopy as a reference standard. Through a survey at study start, study end, and 6-month intervals, readers were asked whether their interpretive preference in clinical practice was to perform a primary 2D, primary 3D, or both 2D and 3D interpretation. Readers were randomly assigned a primary interpretation method (2D or 3D) for each CT colonographic examination. Sensitivity and specificity of each method (primary 2D or 3D), for detecting polyps of 10 mm or larger and 6 mm or larger, based on interpretive preference were estimated by using resampling methods.

RESULTS

Little change was observed in readers' preferences when comparing them at study start and study end, respectively, as follows: primary 2D (eight and seven readers), primary 3D (one and two readers), and both 2D and 3D (six and six readers). Sensitivity and specificity, respectively, for identifying examinations with polyps of 10 mm or larger for readers with a primary 2D preference (n = 1128 examinations) were 0.84 and 0.86, which was not significantly different from 0.84 and 0.83 for readers who preferred 2D and 3D (n = 1025 examinations) or from 0.76 and 0.82 for readers with a primary 3D preference (n = 378 examinations). When performance by using the assigned 2D or 3D method was evaluated on the basis of 2D or 3D preference, there was no difference among those readers by using their preferred versus not preferred method of interpretation. Similarly, no significant difference among readers or preferences was seen when performance was evaluated for detection of polyps of 6 mm or larger.

CONCLUSION

The reader's preference for interpretive method had no effect on CT colonographic performance.

Incremental benefit of computer-aided detection when used as a second and concurrent reader of CT colonographic data: multiobserver study

PURPOSE

To quantify the changes in reader performance levels, if any, during interpretation of computed tomographic (CT) colonographic data when a computer-aided detection (CAD) system is used as a second or concurrent reader.

MATERIALS AND METHODS

After institutional review board approval was obtained, 16 experienced radiologists searched for polyps in 112 patients, 56 of whom had 132 polyps. Each case was interpreted on three separate occasions by using an unassisted (without CAD), second-read CAD, or concurrent CAD reading paradigm. The reading paradigm and case order were randomized, with a minimal interval of 1 month between consecutive interpretations. The readers' findings were compared with the reference-truth interpretation. The mean per-patient sensitivity and mean per-patient specificity with CAD were compared with those achieved with unassisted reading. An increase in per-patient sensitivity was considered to be clinically more important than an equivalent decrease in specificity.

RESULTS

The mean per-patient sensitivity for identification of patients with polyps of any size increased significantly with use of second-read CAD (mean increase, 7.0%; 95% confidence interval [CI]: 4.0%, 9.8%) and concurrent CAD (mean increase, 4.5%; 95% CI: 0.8%, 8.2%). The mean per-patient specificity did not decrease significantly with use of second-read CAD (mean decrease, -2.5%; 95% CI: -5.2%, 0.1%) or concurrent CAD (mean decrease, -2.2%; 95% CI: -4.6%, 0.2%). With analysis restricted to patients with polyps 6 mm or larger, the benefit in sensitivity with second-read CAD remained (mean increase, 7.1%; 95% CI: 3.0%, 11.1%), whereas the increase with concurrent CAD was not significant (mean increase, 4.2%; 95% CI: -0.5%, 8.9%). Use of second-read CAD significantly increased the per-polyp sensitivity for polyps 6 mm or larger (mean increase, 9.0%; 98.3% CI: 4.9%, 12.8%) and polyps 5 mm or smaller (mean increase, 5.9%; 98.3% CI: 3.2%, 9.1%), but use of concurrent CAD increased the per-polyp sensitivity for only those polyps 5 mm or smaller (mean increase, 4.8%; 98.3% CI: 2.2%, 7.9%).

CONCLUSION

Use of second-read CAD significantly improves readers' per-patient and per-polyp detection. Concurrent CAD is less effective.

SUPPLEMENTAL MATERIAL

http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.10100354/-/DC1.

Colorectal polyps: stand-alone performance of computer-aided detection in a large asymptomatic screening population

PURPOSE

To evaluate stand-alone performance of computer-aided detection (CAD) for colorectal polyps of 6 mm or larger at computed tomographic (CT) colonography in a large asymptomatic screening cohort.

MATERIALS AND METHODS

In this retrospective, institutional review board-approved, HIPAA-compliant study, a CAD software system was applied to screening CT colonography in 1638 women and 1408 men (mean age, 56.9 years) evaluated at a single medical center between March 2006 and December 2008. All participants underwent cathartic preparation with stool tagging; electronic cleansing was not used. The reference standard consisted of interpretation by experienced radiologists in all cases. This interpretation was further refined for the subset of cases with positive findings by using subsequent colonoscopic or CT colonographic confirmation, as well as retrospective expert localization of polyps with CT colonography. This test set was not involved in training the CAD system. The Fisher exact test was used to evaluate significance; 95% confidence intervals (CIs) were obtained by using the exact method.

RESULTS

Per-patient CAD sensitivities were 93.8% (350 of 373; 95% CI: 90.9%, 96.1%) and 96.5% (137 of 142; 95% CI: 92.0%, 98.8%) at 6- and 10-mm threshold sizes, respectively. Per-polyp CAD sensitivities for all polyps, regardless of histologic features, were 90.1% (547 of 607; 95% CI: 88.0%, 92.8%) and 96.0% (168 of 175; 95% CI: 91.9%, 98.4%) at 6- and 10-mm threshold sizes, respectively; CAD sensitivities for advanced neoplasia and cancer were 97.0% (128 of 132; 95% CI: 92.4%, 99.2%) and 100% (13 of 13; 95% CI: 79.4%, 100%), respectively. The mean and median false-positive rates were 4.7 and 3 per series, respectively (9.4 and 6 per patient). Among 373 patients with a positive finding at CT colonography, CAD marked an additional 15 polyps of 6 mm or larger, including four large polyps, that were missed at the prospective three-dimensional reading by an expert but were found at subsequent colonoscopy.

CONCLUSION

Stand-alone CAD demonstrated excellent performance for polyp detection in a large screening population, with high sensitivity and an acceptable number of false-positive results.

CT colonography and non-polypoid colorectal neoplasms

Computed tomographic colonography (CTC) has been reported to be as effective as optical colonoscopy in the detection of significant adenomas. However, there are widely conflicting performance data in relation to detection of flat neoplasia. This article describes the potential and limitations of CTC and computer-aided diagnosis in the detection of flat neoplasms.

CT colonography: advanced computer-aided detection scheme utilizing MTANNs for detection of "missed" polyps in a multicenter clinical trial

PURPOSE

The purpose of this study was to develop an advanced computer-aided detection (CAD) scheme utilizing massive-training artificial neural networks (MTANNs) to allow detection of "difficult" polyps in CT colonography (CTC) and to evaluate its performance on false-negative (FN) CTC cases that radiologists "missed" in a multicenter clinical trial.

METHODS

The authors developed an advanced CAD scheme consisting of an initial polyp-detection scheme for identification of polyp candidates and a mixture of expert MTANNs for substantial reduction in false positives (FPs) while maintaining sensitivity. The initial polyp-detection scheme consisted of (1) colon segmentation based on anatomy-based extraction and colon-based analysis and (2) detection of polyp candidates based on a morphologic analysis on the segmented colon. The mixture of expert MTANNs consisted of (1) supervised enhancement of polyps and suppression of various types of nonpolyps, (2) a scoring scheme for converting output voxels into a score for each polyp candidate, and (3) combining scores from multiple MTANNs by the use of a mixing artificial neural network. For testing the advanced CAD scheme, they created a database containing 24 FN cases with 23 polyps (range of 6-15 mm; average of 8 mm) and a mass (35 mm), which were "missed" by radiologists in CTC in the original trial in which 15 institutions participated.

RESULTS

The initial polyp-detection scheme detected 63% (15/24) of the missed polyps with 21.0 (505/24) FPs per patient. The MTANNs removed 76% of the FPs with loss of one true positive; thus, the performance of the advanced CAD scheme was improved to a sensitivity of 58% (14/24) with 8.6 (207/24) FPs per patient, whereas a conventional CAD scheme yielded a sensitivity of 25% at the same FP rate (the difference was statistically significant).

CONCLUSIONS

With the advanced MTANN CAD scheme, 58% of the polyps missed by radiologists in the original trial were detected and with a reasonable number of FPs. The results suggest that the use of an advanced MTANN CAD scheme may potentially enhance the detection of "difficult" polyps.

Improving the accuracy of CTC interpretation: computer-aided detection

Computer-aided polyp detection aims to improve the accuracy of the colonography interpretation. The computer searches the colonic wall to look for polyplike protrusions and presents a list of suspicious areas to a physician for further analysis. Computer-aided polyp detection has developed rapidly in the past decade in the laboratory setting and has sensitivities comparable with those of experts. Computer-aided polyp detection tends to help inexperienced readers more than experienced ones and may also lead to small reductions in specificity. In its currently proposed use as an adjunct to standard image interpretation, computer-aided polyp detection serves as a spellchecker rather than an efficiency enhancer.

Polyp measurement based on CT colonography and colonoscopy: variability and systematic differences

Objective

To assess the variability and systematic differences in polyp measurements on optical colonoscopy and CT colonography.

Materials

Gastroenterologists measured 51 polyps by visual estimation, forceps comparison and linear probe. CT colonography observers randomly assessed polyp size two-dimensionally (abdominal and intermediate window) and three-dimensionally (manually and semi-automatically). Linear mixed models were used to assess the variability and systematic differences between CT colonography and optical colonoscopy techniques.

Results

The variability of forceps and linear probe measurements was comparable and both showed less variability than measurement by visual assessment. Measurements by linear probe were 0.7 mm smaller than measurements by visual assessment or by forceps. The variability of all CT colonography techniques was lower than for measurements by forceps or visual assessment and sometimes lower (only 2D intermediate window and manual 3D) compared with measurements by linear probe. All CT colonography measurements judged polyps to be larger than optical colonoscopy, with differences ranging from 0.7 to 2.3 mm.

Conclusion

A linear probe does not reduce the measurement variability of endoscopists compared with the forceps. Measurement differences between observers on CT colonography were usually smaller than at optical colonoscopy. Polyps appeared larger when using various CT colonography techniques than when measured during optical colonoscopy.

Automated computer-aided stenosis detection at coronary CT angiography: initial experience

OBJECTIVE

To evaluate the performance of a computer-aided algorithm for automated stenosis detection at coronary CT angiography (cCTA).

METHODS

We investigated 59 patients (38 men, mean age 58 +/- 12 years) who underwent cCTA and quantitative coronary angiography (QCA). All cCTA data sets were analyzed using a software algorithm for automated, without human interaction, detection of coronary artery stenosis. The performance of the algorithm for detection of stenosis of 50% or more was compared with QCA.

RESULTS

QCA revealed a total of 38 stenoses of 50% or more of which the algorithm correctly identified 28 (74%). Overall, the automated detection algorithm had 74%/100% sensitivity, 83%/65% specificity, 46%/58% positive predictive value, and 94%/100% negative predictive value for diagnosing stenosis of 50% or more on per-vessel/per-patient analysis, respectively. There were 33 false positive detection marks (average 0.56/patient), of which 19 were associated with stenotic lesions of less than 50% on QCA and 14 were not associated with an atherosclerotic surrogate.

CONCLUSION

Compared with QCA, the automated detection algorithm evaluated has relatively high accuracy for diagnosing significant coronary artery stenosis at cCTA. If used as a second reader, the high negative predictive value may further enhance the confidence of excluding significant stenosis based on a normal or near-normal cCTA study.