1
|
Grudza M, Salinel B, Zeien S, Murphy M, Adkins J, Jensen CT, Bay C, Kodibagkar V, Koo P, Dragovich T, Choti MA, Kundranda M, Syeda-Mahmood T, Wang HZ, Chang J. Methods for improving colorectal cancer annotation efficiency for artificial intelligence-observer training. World J Radiol 2023; 15:359-369. [PMID: 38179201 PMCID: PMC10762523 DOI: 10.4329/wjr.v15.i12.359] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Download PDF] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 10/03/2023] [Revised: 11/13/2023] [Accepted: 12/05/2023] [Indexed: 12/26/2023] Open
Abstract
BACKGROUND Missing occult cancer lesions accounts for the most diagnostic errors in retrospective radiology reviews as early cancer can be small or subtle, making the lesions difficult to detect. Second-observer is the most effective technique for reducing these events and can be economically implemented with the advent of artificial intelligence (AI). AIM To achieve appropriate AI model training, a large annotated dataset is necessary to train the AI models. Our goal in this research is to compare two methods for decreasing the annotation time to establish ground truth: Skip-slice annotation and AI-initiated annotation. METHODS We developed a 2D U-Net as an AI second observer for detecting colorectal cancer (CRC) and an ensemble of 5 differently initiated 2D U-Net for ensemble technique. Each model was trained with 51 cases of annotated CRC computed tomography of the abdomen and pelvis, tested with 7 cases, and validated with 20 cases from The Cancer Imaging Archive cases. The sensitivity, false positives per case, and estimated Dice coefficient were obtained for each method of training. We compared the two methods of annotations and the time reduction associated with the technique. The time differences were tested using Friedman's two-way analysis of variance. RESULTS Sparse annotation significantly reduces the time for annotation particularly skipping 2 slices at a time (P < 0.001). Reduction of up to 2/3 of the annotation does not reduce AI model sensitivity or false positives per case. Although initializing human annotation with AI reduces the annotation time, the reduction is minimal, even when using an ensemble AI to decrease false positives. CONCLUSION Our data support the sparse annotation technique as an efficient technique for reducing the time needed to establish the ground truth.
Collapse
Affiliation(s)
- Matthew Grudza
- School of Biological Health and Systems Engineering, Arizona State University, Tempe, AZ 85287, United States
| | - Brandon Salinel
- Department of Radiology, Banner MD Anderson Cancer Center, Gilbert, AZ 85234, United States
| | - Sarah Zeien
- School of Osteopathic Medicine, A.T. Still University, Mesa, AZ 85206, United States
| | - Matthew Murphy
- School of Osteopathic Medicine, A.T. Still University, Mesa, AZ 85206, United States
| | - Jake Adkins
- Department of Abdominal Imaging, MD Anderson Cancer Center, Houston, TX 77030, United States
| | - Corey T Jensen
- Department of Abdominal Imaging, University Texas MD Anderson Cancer Center, Houston, TX 77030, United States
| | - Curtis Bay
- Department of Interdisciplinary Sciences, A.T. Still University, Mesa, AZ 85206, United States
| | - Vikram Kodibagkar
- School of Biological and Health Systems Engineering, Arizona State University, Tempe, AZ 85287, United States
| | - Phillip Koo
- Department of Radiology, Banner MD Anderson Cancer Center, Gilbert, AZ 85234, United States
| | - Tomislav Dragovich
- Division of Cancer Medicine, Banner MD Anderson Cancer Center, Gilbert, AZ 85234, United States
| | - Michael A Choti
- Department of Surgical Oncology, Banner MD Anderson Cancer Center, Gilbert, AZ 85234, United States
| | - Madappa Kundranda
- Division of Cancer Medicine, Banner MD Anderson Cancer Center, Gilbert, AZ 85234, United States
| | | | - Hong-Zhi Wang
- IBM Almaden Research Center, IBM, San Jose, CA 95120, United States
| | - John Chang
- Department of Radiology, Banner MD Anderson Cancer Center, Gilbert, AZ 85234, United States
| |
Collapse
|
2
|
Zheng S, Schrijvers JJA, Greuter MJW, Kats-Ugurlu G, Lu W, de Bock GH. Effectiveness of Colorectal Cancer (CRC) Screening on All-Cause and CRC-Specific Mortality Reduction: A Systematic Review and Meta-Analysis. Cancers (Basel) 2023; 15:cancers15071948. [PMID: 37046609 PMCID: PMC10093633 DOI: 10.3390/cancers15071948] [Citation(s) in RCA: 2] [Impact Index Per Article: 2.0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/13/2023] [Revised: 03/14/2023] [Accepted: 03/22/2023] [Indexed: 04/14/2023] Open
Abstract
(1) Background: The aim of this study was to pool and compare all-cause and colorectal cancer (CRC) specific mortality reduction of CRC screening in randomized control trials (RCTs) and simulation models, and to determine factors that influence screening effectiveness. (2) Methods: PubMed, Embase, Web of Science and Cochrane library were searched for eligible studies. Multi-use simulation models or RCTs that compared the mortality of CRC screening with no screening in general population were included. CRC-specific and all-cause mortality rate ratios and 95% confidence intervals were calculated by a bivariate random model. (3) Results: 10 RCTs and 47 model studies were retrieved. The pooled CRC-specific mortality rate ratios in RCTs were 0.88 (0.80, 0.96) and 0.76 (0.68, 0.84) for guaiac-based fecal occult blood tests (gFOBT) and single flexible sigmoidoscopy (FS) screening, respectively. For the model studies, the rate ratios were 0.45 (0.39, 0.51) for biennial fecal immunochemical tests (FIT), 0.31 (0.28, 0.34) for biennial gFOBT, 0.61 (0.53, 0.72) for single FS, 0.27 (0.21, 0.35) for 10-yearly colonoscopy, and 0.35 (0.29, 0.42) for 5-yearly FS. The CRC-specific mortality reduction of gFOBT increased with higher adherence in both studies (RCT: 0.78 (0.68, 0.89) vs. 0.92 (0.87, 0.98), model: 0.30 (0.28, 0.33) vs. 0.92 (0.51, 1.63)). Model studies showed a 0.62-1.1% all-cause mortality reduction with single FS screening. (4) Conclusions: Based on RCTs and model studies, biennial FIT/gFOBT, single and 5-yearly FS, and 10-yearly colonoscopy screening significantly reduces CRC-specific mortality. The model estimates are much higher than in RCTs, because the simulated biennial gFOBT assumes higher adherence. The effectiveness of screening increases at younger screening initiation ages and higher adherences.
Collapse
Affiliation(s)
- Senshuang Zheng
- Medical Center Groningen, Department of Epidemiology, University of Groningen, 9700 RB Groningen, The Netherlands
| | - Jelle J A Schrijvers
- Medical Center Groningen, Department of Epidemiology, University of Groningen, 9700 RB Groningen, The Netherlands
| | - Marcel J W Greuter
- Medical Center Groningen, Department of Radiology, University of Groningen, 9700 RB Groningen, The Netherlands
- Robotics and Mechatronics (RaM) Group, Technical Medical Centre, Faculty of Electrical Engineering Mathematics and Computer Science, University of Twente, 7522 NH Enschede, The Netherlands
| | - Gürsah Kats-Ugurlu
- Medical Center Groningen, Department of Pathology, University of Groningen, 9700 RB Groningen, The Netherlands
| | - Wenli Lu
- Department of Epidemiology and Health Statistics, Tianjin Medical University, Tianjin 300070, China
| | - Geertruida H de Bock
- Medical Center Groningen, Department of Epidemiology, University of Groningen, 9700 RB Groningen, The Netherlands
| |
Collapse
|
3
|
Ebner D, Kisiel J, Barnieh L, Sharma R, Smith NJ, Estes C, Vahdat V, Ozbay AB, Limburg P, Fendrick AM. The cost-effectiveness of non-invasive stool-based colorectal cancer screening offerings from age 45 for a commercial and medicare population. J Med Econ 2023; 26:1219-1226. [PMID: 37752872 DOI: 10.1080/13696998.2023.2260681] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Received: 07/06/2023] [Accepted: 09/15/2023] [Indexed: 09/28/2023]
Abstract
AIM The United States Preventive Services Taskforce (USPSTF) recently recommended lowering the age for average-risk colorectal cancer (CRC) screening from 50 to 45 years. While initiating screening at age 45 versus 50 provides a greater opportunity for CRC early detection and prevention, the full profile of benefits, risks, and cost-effectiveness of expanding the screen-eligible population requires further evaluation. MATERIALS AND METHODS The costs and clinical outcomes for screening at age 45 for triennial multi-target stool DNA [mt-sDNA], and other non-invasive stool-based modalities (annual fecal immunochemical test [FIT] and annual fecal-occult blood test [FOBT]), were estimated using the validated CRC-AIM microsimulation model over a lifetime horizon. Test sensitivity and specificity inputs were based on 2021 USPSTF modeling analyses; adherence rates were based on published real-world data and the costs of the screening test, follow-up colonoscopies, complications, and CRC care were included. Outcomes are reported from the perspective of a United States payer as clinical, life-years gained (LYG), and incremental cost-effectiveness ratio (ICER); stool-based and follow-up colonoscopy adherence ranges were explored in one-way, probabilistic and threshold analyses. RESULTS When compared to initiation of CRC screening at age 45 versus 50, all modalities reduced both the incidence of and mortality from CRC and increased LYG. Initiating CRC screening at age 45 was cost-effective with an ICER of $59,816 and $35,857 per quality-adjusted life year (QALY) for mt-sDNA versus FIT and FOBT, respectively. In the threshold analyses, at equivalent rates to stool-based screening, mt-sDNA was always cost-effective at a willingness-to-pay threshold of $100,000 per QALY versus FIT and FOBT. CONCLUSIONS Initiating average-risk CRC screening at age 45 instead of age 50 increases the estimated clinical benefit by reducing disease burden while remaining cost-effective. Among stool-based screening modalities, mt-sDNA provides the most clinical benefit in a Commercial and Medicare population.
Collapse
Affiliation(s)
- Derek Ebner
- Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, MN, USA
| | - John Kisiel
- Division of Gastroenterology and Hepatology, Mayo Clinic, Rochester, MN, USA
| | | | | | | | | | | | | | | | - A Mark Fendrick
- Center for Value Based Insurance Design, University of Michigan, Ann Arbor, MI, USA
| |
Collapse
|
4
|
Tzovaras AA, Karatrasoglou E, Talagani S, Perdikari K, Veros C, Nikolaou M. Implications of the reduction in prostate cancer screening using prostate-specific antigen. Curr Med Res Opin 2022; 38:1383. [PMID: 35369834 DOI: 10.1080/03007995.2022.2061708] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [MESH Headings] [Track Full Text] [Journal Information] [Submit a Manuscript] [Subscribe] [Scholar Register] [Indexed: 11/03/2022]
Affiliation(s)
- Alexandros A Tzovaras
- First Medical Oncology Clinic, Saint-Savvas Cancer Hospital L. Alexandras 171, Athens 11522, Greece
| | | | - Sofia Talagani
- First Medical Oncology Clinic, Saint-Savvas Cancer Hospital, Athens, Greece
| | | | - Christos Veros
- First Medical Oncology Clinic, Saint-Savvas Cancer Hospital, Athens, Greece
| | - Michail Nikolaou
- First Medical Oncology Clinic, Saint-Savvas Cancer Hospital, Athens, Greece
| |
Collapse
|