Table 1.

Summary of recent studies on the implementation of computer-aided detection in clinical settings for breast cancer detection.

Author, year, country	Methods	Objectives	Results
Keen et al, 2018, United States [10]	Telephone surveys (400 digital mammography practices)	To assess whether CADa use by digital mammography practices decreased from 2008 to 2016	CAD use remained stable from 2008 to 2016 at US digital mammography practices (91.4% in 2008, 90.2% in 2011, and 92.3% in 2016).
Fenton et al, 2013, United States [22]	Retrospective cohort study of Medicare enrollees from the Surveillance, Epidemiology, and End Results Medicare database (409,459 mammograms and 163,099 women)	To study the relationship between CAD use and DCISb incidence and invasive breast cancer	CAD prevalence increased from 3.6% to 60.5% from 2001 to 2006, respectively. CAD use was linked to greater DCIS incidence. There was no difference in invasive breast cancer incidence; however, invasive breast cancer at earlier stages (I to II vs III to IV) was diagnosed.
Killelea et al, 2014, United States [23]	Retrospective cohort study of Medicare enrollees from the Surveillance, Epidemiology, and End Results Medicare database 2001-2002 (n=137,150) and 2008-2009 (n=133,097)	To evaluate the impact of CAD on screening-related cost and outcomes	CAD use increased from 3.2% to 33.1% from 2001-2002 to 2008-2009, respectively; however, a clinically significant change in stage at diagnosis was not observed.
Rao et al, 2010, United States [24]	Retrospective analysis of nationwide Medicare Part B fee-for-service databases from 2004 to 2008	To compare mammography procedure volumes and CAD use for (1) screening vs diagnostic mammography and (2) hospital facilities vs private offices	CAD was used for 74% of screening mammograms and 50% of diagnostic mammograms by 2008. CAD was used for 70% of hospital-based and 81% of private office-based screening mammograms.
Onega et al, 2010, United States [25]	Cross-sectional survey on the use and perceptions of CAD and double reading by radiologists (n=257)	To examine (1) the rates of CAD and double reading use for mammography interpretation and (2) the perceptions of CAD in comparison to double reading for mammography interpretation	More radiologists perceived that double reading improved cancer detection rates over CAD (74% vs 55% respectively). More than 75% use CAD for some screening mammography interpretation. 72% do not use double reading for screening mammograms.
Tchou et al, 2010, United States [26]	Prospective observational study of radiologists interpreting images with and without CAD (5 radiologists and 267 cases)	To study the effect of CAD on (1) interpretation time for reviewing CAD images, (2) recall rates, and (3) confidence levels	Use of CAD to interpret mammographic images resulted in a 19% or 23 second mean increase in interpretation time and 11% increase in recall rates. Confidence levels of radiologists were altered in 22% of cases: increased confidence in 14% and decreased confidence in 8%.
Benedikt et al, 2018, United States [27]	Prospective study multireader multicase crossover design of images (20 radiologists and 240 cases)	To compare reading time and performance with and without CAD, with concurrent use of DBTc	Concurrent use of CAD with DBT resulted in 29.2% faster reading time while maintaining reader interpretation performance.
Guerriero et al, 2011, United Kingdom [14]	Cost-effectiveness analysis (n=31,057)	To study the cost-effectiveness of single reading plus CAD versus double reading for women having routine screening across low-, average-, and high-volume units	Single reader with CAD is unlikely to be cost-effective, and savings from reading time would be offset by staff training Purchase, upgrading, and maintenance costs involved. Increased cost of assessment, although the model is sensitive to parameters that could change
Sato et al, 2012, Japan [15]	Cost-effective analysis using ICERd ratio	To examine the cost-effectiveness of double reading by two readers versus single reading with CAD	Single reading with CAD for mammography screening is more cost-effective than double reading; results are sensitive to the number of examinees.

^aCAD: computer-aided detection.

^bDCIS: ductal carcinoma in situ.

^cDBT: digital breast tomosynthesis.

^dICER: incremental cost-effectiveness ratio.