BJR
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
British Journal of Radiology (2005) 78, S57-S62
© 2005 British Institute of Radiology
doi: 10.1259/bjr/25777270
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bogoni, L
Right arrow Articles by Macari, M
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bogoni, L
Right arrow Articles by Macari, M

Paper

Computer-aided detection (CAD) for CT colonography: a tool to address a growing need

L Bogoni, PhD 1 P Cathier, PhD 1 M Dundar, PhD 1 A Jerebko, PhD 1 S Lakare, PhD 1 J Liang, PhD 1 S Periaswamy, PhD 1 M E Baker, MD 2 and M Macari, MD 3

1 Computer Aided Diagnosis and Therapy, Siemens Medical Solutions, Malven, PA, 2 Cleveland Clinic Foundation, Cleveland, OH and 3 NYU Medical Center, New York, NY, USA


   Abstract
Top
 Abstract
Introduction
 CTC in studies and...
 Computer-aided detection (CAD)
 Colon CAD: study and...
 Discussion
 Conclusions
 References
 
Colorectal cancer is the third most common cancer in both men and women. It is estimated that in 2004, nearly 147 000 cases of colon and rectal cancer will be diagnosed in the USA, and approximately 57 000 people would die from the disease; however, only 44% of the eligible population undergoes any type of colorectal cancer screening. Many reasons have been identified for non-compliance, with key ones being patient comfort, bowel preparation and cost. Virtual colonoscopy derived from computed tomography (CT) images is gaining broader acceptance as a screening method for colorectal neoplasia. Our research suggests that computer-aided detection (CAD) as a second reader has great potential in improving polyp detection. The ColonCAD prototype presented in this paper was developed and tested on cases representative of the variability and quality in true clinical practice. Results of this study with 150 patients demonstrate that: the developed algorithm generalises well: the sensitivity for polyps =>ge;6 mm is on average 90%; and the median false positive rate is a manageable 3 per volume.


   Introduction
Top
 Abstract
 Introduction
CTC in studies and...
 Computer-aided detection (CAD)
 Colon CAD: study and...
 Discussion
 Conclusions
 References
 
Colorectal cancer (CRC) is the third most common cancer in both men and women. It is estimated [1] that in 2004, nearly 147 000 cases of colon and rectal cancer will be diagnosed in the USA, and more than 56 730 people would die from colon cancer, accounting for approximately 11% of all cancer deaths. Table 1Go illustrates the relationship between early detection of colon cancer and the 5-year survival rate. In particular, since it is known that in over 90% of cases the progression stage for colon cancer is from local (polyp adenomas) to advanced stages (colorectal cancer), it is critical that major efforts be devoted to screening of colon cancer and removal of lesions (polyps) when still in a early stage of the disease. In Winawer et al [2], a guideline for CRC screening is presented, and a guideline for patient management based on CT colonography (CTC) is given in Macari et al [3].


View this table:
[in this window]
[in a new window]
  		Table 1. Colon cancer staging and survival rate and distribution of staging at time of diagnosis. Information integrated colon and rectal cancer from the Cancer Statistics 2004 [1] clearly illustrates the importance of early diagnosis and treatment of colon cancer

 
As evidence of the impact of removing lesions, a study by Citarda et al [4] on 1693 patients, followed over a 10-year period, demonstrated that colonoscopic polypectomy substantially reduced the incidence of CRC in the cohort compared with that expected in the general population. While there is wide consensus that screening patients [5] is effective in decreasing advanced disease, only 44% of the eligible population undergoes any type of CRC screening [6]. Many reasons have been identified for non-compliance, the key ones being patient comfort, bowel preparation and cost [7]. As CTC is gaining broader acceptance as a screening method for colorectal neoplasia [8, 9], results presented in this paper suggest that, in the rather near future, computer-aided detection (CAD) could be employed as a second reader to aid in polyp detection.


   CTC in studies and clinical practice
Top
 Abstract
 Introduction
 CTC in studies and...
Computer-aided detection (CAD)
 Colon CAD: study and...
 Discussion
 Conclusions
 References
 
Virtual colonography was introduced as a new practice by Vining [10] in 1994. Since then, this technique has received increasing attention and has continuously improved with regard to technical protocols. The sensitivity of CTC depends on many technical factors such as patient preparation, colon distension and image quality, as well as the reader's experience.

Studies over the past 5 years have demonstrated that the sensitivity for small polyps (<6 mm) is relatively low, while sensitivity for medium sized polyps (≥6 mm to <10 mm) is between 70% and 90%, and large polyps (>10 mm) can be detected with 90–100% sensitivity (Table 2Go).


View this table:
[in this window]
[in a new window]
  		Table 2. CT colonography (CTC) studies showing the per/polyp sensitivity across different polyp categories. The sensitivity for medium sized and large polyps are the most important since these have strongest clinical relevance

 
Recently, a meta-analysis study by Sosna et al [9] analysed 14 studies with a combined pool of 1324 patients and 1411 polyps. The study reported that the per-polyp sensitivity of CTC was 81% for large polyps, 62% for medium sized polyps and 43% for small polyps.

These numbers indicate that the sensitivity of CTC correlates well with the clinical relevance of the lesions. In addition, the likelihood of malignancy for small polyps was shown to be less than 0.1% [15]. These facts suggest that a CAD system, whilst attempting to achieve high overall sensitivity, would be very well served if it could provide the highest level of sensitivity for the middle to large polyps, thus reflecting the clinical relevance based on size as well as experts' performance in the middle to large sizes.

These studies provide evidence for the future potential of CTC. However, given the large size of the patient population who could benefit from the procedure, it becomes clear that CTC would have to become a ubiquitous practice. With increasing number of CTC procedures being performed at an increasing number of clinical sites, we see a growing demand for CAD products that can help as a second reader in the detection of colon polyps.


   Computer-aided detection (CAD)
Top
 Abstract
 Introduction
 CTC in studies and...
 Computer-aided detection (CAD)
Colon CAD: study and...
 Discussion
 Conclusions
 References
 
The concept of CAD was introduced with mammography CAD as its first application. Over the last few years, a huge number of mammography procedures have been performed with CAD as a second reader. The role of CAD is to alert the radiologist to "regions of interest" that might have been overlooked in the initial read.

The CAD market is developing at a rapid pace and new CAD applications have emerged, such as detection of lung cancer with chest radiography and detection of lung nodules with chest CT. In addition to lung CAD applications, there is increasing interest in the use of CAD technologies for the detection of colon polyps. This can be characterized by the following interrelated needs.

  1. Population-wide: to meet the large demand and to enable effective management of the disease, as discussed earlier.
  2. Clinical: owing to the nature of the colon's anatomy: subject to deformation, tortuosity, etc.; variability of the preparation; distension, insufflation and cleanliness; and the difficulty of detecting of polyps even when examining both prone and supine, since polyps may be visible in only one of the two volumes.
  3. Information and data management: new powerful CT systems have increased spatial resolution (up to 0.4 mm with Siemens Sensation 64) for better diagnostic evaluations but have also increased the number of slices acquired (up to 1200 per volume for a routine abdominal scan).

Therefore, a CAD system must be able to handle the evolving technology, provide better means to visualize and analyse data, and improve physician's sensitivity in the detection of clinically relevant lesions. Whilst visualization and analysis are key to enabling a physician to deliberate on the nature of lesions, the ability to automatically detect lesions (polyps) is where CAD's impact can be most significant. Two types of study designs can be used to determine the sensitivity of a CAD system.

Most studies published so far followed a "comparative" design with the objective of comparing the sensitivity of expert readers with the sensitivity of CAD. The results we present fit in this category.


   Colon CAD: study and performance
Top
 Abstract
 Introduction
 CTC in studies and...
 Computer-aided detection (CAD)
 Colon CAD: study and...
Discussion
 Conclusions
 References
 
The performance of the ColonCAD prototype was developed as part of a study involving data sets obtained from two sites, New York University Medical Center (NYU) and the Cleveland Clinic Foundation (CCF). Both sites granted Independent Review Board approvals and all cases were de-identified (all patient identification information was removed) prior to their transfer to Siemens.

Methods and materials
The database consisted of 150 data sets, 292 volumes from high resolution CT scanners. These included patients with polyps (positive cases; n=64) and patients without polyps (negative cases; n=86). The positive cases include a total of 92 unique polyps greater that 3.0 mm. These cases were partitioned into a working set (training set) and an unseen set (test set). The sensitivity and specificity (number of false positives) of CAD as a tool to aid in polyp detection was established with respect to radiologists' findings on CTC confirmed by concurrent fibre optic colonoscopy. The locations and dimensions of the lesions were then used in subsequent stages to automatically compute sensitivity and specificity with respect to polyp size. The patient protocols and acquisition parameters are shown in Table 3GoGo.


View this table:
[in this window]
[in a new window]
  		Table 3. Patient preparation protocols and acquisition parameters for the data sets from New York University Medical Center (NYU) and Cleveland Clinic Foundation (CCF). Notice that the data from CCF were obtained from two different systems, a Siemens Sensation 16 (CCF1) and a Siemens Volume Zoom (CCF2)

Patient preparation protocol

 

View this table:
[in this window]
[in a new window]
  		Acquisition Parameters

 
Our proprietary ColonCAD algorithm (patents pending) included the following phases: data pre-processing, candidate generation, feature extraction and classification. During the data pre-processing phase, the colon is segmented, and loci of detection are identified in the candidate generation phase. These are sequentially processed during the following phase in which multiple features are extracted. The features are based on tissue intensity, volumetric and surface shape as well as texture characteristics. Each candidate, uniquely identified, and the associated features are then fed to a classifier. Candidates are then evaluated and labelled as potential polyps. The running time on a single volume, 600 slices with 512 x 512 axial resolution, was on average 4 min on a Pentium IV 3.06GHz dual-processor machine with 2 GB of memory.

To automatically process all the cases, a flexible framework was developed that would allow loading the cases (as DICOM images) and would push them through the various stages of the algorithm, outputting intermediate results. The modularity of the system has allowed the effectiveness of each component to be assessed and hence both to be improved and independently refined.

Feature selection and classification
Training and test data
The 150 data sets were randomly partitioned into two groups: training set (n=88) and test set (n=62). The test set was sequestered and used only to evaluate the performance of the final prototype. To automate the training and verification process, a database was developed to allow the software to automatically query the database and provide feedback as to whether the finding was a polyp or non-polyp and in the case of polyps, also to obtain its size. The training set was used to design the classifier and to automatically select the relevant features, as described below.

Feature selection
The feature selection stage is a key component of our approach. We use the "wrapper" method for feature selection [17], in which the classifier decides which features are useful. Procedurally, the classifier is run iteratively on the training set using different feature sets, during each iteration, one or more features are added, until the cross-validation error no longer improves.

Results on training group
The prototype's performance was evaluated on the 88 cases in the training set using leave one patient out (LOPO) cross-validation. In this scheme, both the supine and prone views of a case from the training set were left out. The classifier is trained using the volumes from the remaining 87 (i.e. 88–1) cases, and tested on both volumes of the "left out" case. This process was repeated 88 times, leaving out each of the 88 cases in the training set, and the resulting testing errors were averaged out to determine the LOPO error.

Results on test group
A classifier was trained on all 88 cases in the training set using the 15 features selected in the "Feature Selection" phase. Only after this classifier was frozen was the test group of 62 cases released for evaluation. This provided a very accurate estimate of prototype performance on completely unseen data – the only true test for a classification-based system. In the computation of sensitivity, a polyp was considered as "found" if it was detected in at least one of the volumes (supine or prone) from the same patient.

Results
In this section we review the data characteristics and specific results for the training set and the test set. These are also summarized in Table 4Go. Figure 1Go shows the result of detection of a potential lesion by the ColonCAD prototype as incorporated in the syngo® colonography application.


View this table:
[in this window]
[in a new window]
  		Table 4. Computer-aided detection (CAD) performance. The table summarizes the performance of the ColonCAD prototype as indicated in terms of the two phases: candidate generation and classification. The first column indicates the number of patients, polyps and false positives (fp). The first row characterizes the training phase, while the second row captures the testing phase. For each of the polyp categories (small, medium, large), the sensitivity is expressed first as a fraction of polyps found over the total number of polyps, and next as the percentage. In the first row of the first column from the major column labelled candidate generation (CG), the 16/19 indicates that, of the 88 cases used in the training phase, there were 19 small polyps of which 3 were missed during the CG phase. In the corresponding classification phase, the entry (16–4)/19 shows that 4 small polyps were additionally missed during this phase. The combined sensitivity for middle sized and large polyps are highlighted. Further analysis is presented in the text of the document

 


View larger version (107K):
[in this window]
[in a new window]
  		Figure 1. The ColonCAD prototype as integrated in the syngo® colonography application. The top two quadrants show a coronal reconstruction and an axial slice from CT images displayed in multiplanar reconstruction (MPR) format. The bottom-left quadrant shows a rendered global view of the colon (barium enema rendering) with some labels illustrating the location of findings. The bottom-right quadrant shows the endo-view rendering with a computer-aided detection (CAD) marked finding. This finding is labelled "c9a", where the prefix "c" differentiates it from the physician's findings.

 
Results on the training set
Patient and polyp information.
There were 88 cases with 171 volumes (some cases did not have both prone and supine studies). A total of 53 unique polyps were identified in this set: 19 small (less than 6 mm), 25 mid-size (6–9 mm) and 9 large (10–20 mm).

Candidate generation.
The candidate generation stage generated an average of 48.2 candidates per volume while missing 3 small sized polyps.

Classifier results.
We obtained a median false positive (fp) rate of 3.5 per volume. The sensitivities obtained for different ranges of polyps are as follows: small=63.1%, mid-size = 92.0%, large = 88.9%, overall = 81.1% and overall =>ge;5 mm = 91.2%.

Results on the test set (previously unseen data)
Patient and polyp information.
There were 62 cases with 121 volumes. A total of 39 unique polyps were identified in this set: 18 small, 11 mid-size and 10 large.

Candidate generation.
The candidate generation stage generated an average of 51.4 candidates per volume while missing 5 small polyps and 1 medium polyp.

Classifier results.
We obtained a median false positive rate of 3 per volume. The sensitivities obtained for different ranges of polyps are as follows: small = 66.7%, mid-size = 81.8%, large = 100%, overall = 82.1% and overall =>ge;5 mm = 90.5%.


   Discussion
Top
 Abstract
 Introduction
 CTC in studies and...
 Computer-aided detection (CAD)
 Colon CAD: study and...
 Discussion
Conclusions
 References
 
The ColonCAD results of 90% sensitivity for detection of medium–large colon polyps compares favourably both with detection rates in published CTC studies and with published results from other CAD systems (Table 5Go).


View this table:
[in this window]
[in a new window]
  		Table 5. Computer-aided detection (CAD) Studies. Development of colon CAD systems has only begun recently and to date only a limited number of studies have begun to assess CAD sensitivity beyond just proving its feasibility on a handful of patients. Some of the tabulated values were estimated from the information available in the literature; information for entries marked as unk. (unknown) was not available

 
Results shown in Table 5Go demonstrate that CAD technology is able to automatically detect colon polyps of clinical significance with a high level of sensitivity. A direct comparison across the different CAD systems is, however, difficult given the variability in validation methods, patient population, colon preparation, data acquisition protocols and quality, as well as how the information is reported on sensitivity (per patient, per polyp, per volume) and computation of false positives.

Training and testing is of paramount importance to ensure that a CAD system works well across a variety of patients (from different sites), including patients with suboptimally prepared colon, motion artefacts caused by bowel movements or interruption of the breath-hold during scanning. During development of our ColonCAD prototype, we paid attention to these requirements and kept all cases in both training and test sets so as to comply with our goal to develop a CAD system that would (a) closely resemble the type of cases reviewed in clinical settings and (b) generalize well (be robust to future testing when presented with new cases).

CAD has the potential to increase sensitivity, however as a second reader it will add to the reading time. Therefore, workflow is very important so that CAD marks can be reviewed efficiently and the number of false positive detection marks remains at a manageable level.

Our ColonCAD prototype median of false positive detections has remained at 3 per volume and we believe that future advancement of the technology can further reduce this number. While some false positive detections can be easily dismissed owing to their anatomical location, other marks, such as labelling of faecal matter, may not be easily differentiated at first look. In the case of large structures, analysis of its density and morphology can help to dismiss the detection. Additionally, motility [24] may be used to weed out these findings and some cases may be rejected by a CAD system by the use of automatic prone–supine registration [23]. However, since it was shown that local motion in filling defects is also associated with lesions, it should not be assumed to be indicative of faecal material [23]. Thus, integration of other means, such as tagging or electronic cleansing, may be needed to handle source of false positives.

Since ColonCAD has shown good results in "comparative" studies, the next step is to demonstrate value in "additive" studies in which the CAD prototype functions as a second reader.


   Conclusions
Top
 Abstract
 Introduction
 CTC in studies and...
 Computer-aided detection (CAD)
 Colon CAD: study and...
 Discussion
 Conclusions
References
 
The developed ColonCAD prototype has focused on the detection of polyps of sizes up to and including 20 mm. The results have demonstrated that:

  1. the developed ColonCAD algorithm generalizes well when applied on a completely unseen test set;
  2. the system performs well on routine clinical cases – these included some cases that had poorly distended and under insufflated colons, artefacts due to motion and prosthetic implants;
  3. the sensitivity for middle to large sized polyps is on average 90%;
  4. while the false positive rates can be improved, they have remained at a median of 3 per volume;
  5. the reported sensitivity for the ColonCAD algorithm is well in the range of the reported sensitivities of expert radiologists.

Additionally, three polyps larger than 6 mm and one polyp smaller than 5 mm, which were detected by the ColonCAD prototype, were only found retrospectively by the radiologist following fibre optic colonoscopy (OC). The ColonCAD prototype also detected two polyps larger than 6 mm that were missed by OC but found by the radiologist, and later found in OC. These observations further demonstrate the value of this CAD prototype. While the data sets also included 12 masses, these have not been incorporated as part of the statistics reported.


   References
Top
 Abstract
 Introduction
 CTC in studies and...
 Computer-aided detection (CAD)
 Colon CAD: study and...
 Discussion
 Conclusions
 References
 

  1. Jemal A, Tiwari RC, Murray T, Ghafoor A, Saumuels A, Ward E, et al. Cancer Statistics 2004. CA Cancer J Clin 2004;54:8–29.[Abstract/Free Full Text]
  2. Winawer S, Fletcher R, Rex D, Bond J, Burt R, Ferrucci J, et al. Colorectal cancer screening and surveillance: clinical guidelines and rationale – update based on new evidence. Gastroenterology 2003;124:544–60.[CrossRef][Medline]
  3. Macari M, Bini EJ, Jacobs SL, Lui YW, Laks S, Milano A, et al. Significance of missed polyps at CT colonography. AJR Am J Roentgenol 2004;183:127–34.[Abstract/Free Full Text]
  4. Citarda F, Tomaselli G, Capocaccia R, Barcherini S, Crespi M. Efficacy in standard clinical practice of colonoscopic polypectomy in reducing colorectal cancer incidence. Gut 2001;48:812–5.[Abstract/Free Full Text]
  5. Morrin MM, LaMont JT. Screening virtual colonoscopy — ready for prime time? N Engl J Med 2003;349:2261–4.[Free Full Text]
  6. Anon. Trends in screening for colorectal cancer — United States, 1997 and 1999. MMWR Morb Mortal Wkly Rep 2001;50:162–6.[Medline]
  7. Zack DL, DiBaise JK, Quigley EM, et al. Colorectal cancer screening compliance by medicine residents: perceived and actual. Am J Gastroenterol 2001;96:3002–8.
  8. Pickhardt PJ, Choi JR, Hwang I, Butler JA, Puckett JL, Hildebrand HA. Computer tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med 2003;349:2191–200.[Abstract/Free Full Text]
  9. Sosna J, Morrin MM, Kruskal JB, Lavin PT, Rosen MP, Raptopoulos V. CT colonography of colorectal polyps: a metaanalysis. AJR 2003;181:1593–8.[Abstract/Free Full Text]
  10. Vining DJ, Gelfand DW, Bechtold RE, Scharling ES, Grishaw EK, Shifrin RY. Technical feasibility of colon imaging with helical CT and virtual reality. AJR Am J Roentgenol 1994;62(Suppl.):104.
  11. Fenlon HM, Nunes DP, Schroy PCII, et al. A comparison of virtual and conventional colonoscopy for the detection of colorectal polyps. N Engl J Med 1999;341:1496–503.[Abstract/Free Full Text]
  12. Yee J, Geetanjali AA, Hung RK, Steinauer-Gebauer AM, Wall AD, McQuaid KR. Colorectal neoplasia: performance characteristics of CT colonography for detection in 300 patients. Radiology 2001;219:685–92.[Abstract/Free Full Text]
  13. Macari M, Bini EJ, Xue X, Milano A, Katz SS, Resnick D, et al. Colorectal neoplasms: prospective comparison of thin-section low-dose multi-detector row CT colonography and conventional colonoscopy for detection. Radiology 2002;224:383–92.[Abstract/Free Full Text]
  14. Iannaccone R, Laghi A, Catalano C, Brink JA, Mangiapane F, Trenna S, et al. Detection of colorectal lesions: lower-dose multi-detector row helical CT colonography compared with conventional colonoscopy. Radiology 2003;229:775–81.[Abstract/Free Full Text]
  15. Waye JD, Lewis BS, Frankel A, Geller SA. Small colon polyps. Am J Gastroenterol 1988;83:120–2.[Medline]
  16. Okamura A, Dachman AH, Parsad N, Nappi J, Yoshida H. Evaluation of the effect of CAD on observers' performance in detection of polyps in CT colonography. In: 6th International Workshop on Computer Aided Diagnosis. CARS 2004;1268:989–92.
  17. John GH, Kohavi R, Pfleger K. Irrelevant features and the subset selection problem. In: Proceedings of the 11th International Conference on Machine Learning, California: Morgan Kaufmann, 1994:121–9.
  18. Jerebko AK, Malley JD, Franaszek M, Summers RM. Multinetwork classification scheme for detection of colonic polyps in CT colonography data sets. Acad Radiol 2003;10:154–60.
  19. Yoshida H, Nappi J, MacEneaney P, Rubin DT, Dachman AH. Computer-aided diagnosis scheme for detection of polyps at CT colonography. Radiographics 2002;22:963–79.[Abstract/Free Full Text]
  20. Gokturk SB, Tomasi C, Acar B, Beaulieu CF, Paik DS, Jeffrey RB, et al. A statistical 3-D pattern processing method for computer-aided detection of polyps in CT colonography. IEEE Trans Med Imaging 2001;20:1251–60.[CrossRef][Medline]
  21. Li H, Santago P. A practical automated polyp detection scheme for CT colonography. SPIE Med Imag., 2004;5372:487–95.
  22. Kiss G, Van Cleynenbreugel J, Suetens P, Marchal G. Computer aided diagnosis for CT colonography via slope density functions. MICCAI 2003;2878:746–53.
  23. Nappi J, Frimmel H, Okamura A, Dachman AH, Yoshida H. Region-based supine–pro correspondence for reduction of false positives in CAD of CT colonography. CARS 2004;1268:993–8.
  24. Laks S, Macari M, Bini EJ. Positional change in colon polyps at CT colonography. Radiology 2004;231:761–6.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 RadiologyHome page
D. Hock, R. Ouhadi, R. Materne, A.-S. Aouchria, I. Mancini, T. Broussaud, P. Magotteaux, and A. Nchimi
Virtual Dissection CT Colonography: Evaluation of Learning Curves and Reading Times with and without Computer-aided Detection
Radiology, September 1, 2008; 248(3): 860 - 868.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Roentgenol.Home page
R. M. Summers, L. R. Handwerker, P. J. Pickhardt, R. L. Van Uitert, K. K. Deshpande, S. Yeshwant, J. Yao, and M. Franaszek
Performance of a Previously Validated CT Colonography Computer-Aided Detection System in a New Patient Population
Am. J. Roentgenol., July 1, 2008; 191(1): 168 - 174.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
S. A. Taylor, R. Greenhalgh, R. Ilangovan, E. Tam, V. A. Sahni, D. Burling, J. Zhang, P. Bassett, P. J. Pickhardt, and S. Halligan
CT Colonography and Computer-aided Detection: Effect of False-Positive Results on Reader Specificity and Reading Efficiency in a Low-Prevalence Screening Population
Radiology, April 1, 2008; 247(1): 133 - 140.
[Abstract] [Full Text] [PDF]

Home page
 Br. J. Radiol.Home page
S A TAYLOR, D BURLING, M RODDIE, L HONEYFIELD, J MCQUILLAN, P BASSETT, and S HALLIGAN
Computer-aided detection for CT colonography: incremental benefit of observer training
Br. J. Radiol., March 1, 2008; 81(963): 180 - 186.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
C. Robinson, S. Halligan, S. A. Taylor, S. Mallett, and D. G. Altman
CT Colonography: A Systematic Review of Standard of Reporting for Studies of Computer-aided Detection
Radiology, February 1, 2008; 246(2): 426 - 433.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
M. E. Baker, L. Bogoni, N. A. Obuchowski, C. Dass, R. M. Kendzierski, E. M. Remer, D. M. Einstein, P. Cathier, A. Jerebko, S. Lakare, et al.
Computer-aided Detection of Colorectal Polyps: Can It Improve Sensitivity of Less-Experienced Readers? Preliminary Findings
Radiology, October 1, 2007; 245(1): 140 - 149.
[Abstract] [Full Text] [PDF]

Home page
 Am. J. Roentgenol.Home page
S. Jensch, R. E. van Gelder, J. Florie, M. A. T.-d. Graaf, J. V. Lobe, P. M. M. Bossuyt, S. Bipat, C. Y. Nio, and J. Stoker
Performance of Radiographers in the Evaluation of CT Colonographic Images
Am. J. Roentgenol., March 1, 2007; 188(3): W249 - W255.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Bogoni, L
Right arrow Articles by Macari, M
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Bogoni, L
Right arrow Articles by Macari, M

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
BJR DMFR IMAGING  ALL BIR JOURNALS