Automatic Scan Range Delimitation in Chest CT Using Deep                     Learning

Demircioğlu, Aydın; Kim, Moon-Sung; Stein, Magdalena Charis; Guberina, Nika; Umutlu, Lale; Naßenstein, Kai

doi:10.1148/ryai.2021200211

Cited by 13 publications

(10 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As opposed to DL algorithms reported in the abovementioned studies [ 7 ], the model in our study detects under-scanning (missed scan coverage) in addition to over-scanning. To our best knowledge, no prior studies have reported on the use of DL for identifying missed anatomic coverage, which can lead to either incomplete diagnostic evaluation or trigger additional scanning over the missed anatomic region.…”

Section: Discussionmentioning

confidence: 94%

“…The authors reported an over-scanning rate of 22.6% in 1000 chest CT examinations assessed with their algorithm. Another study from Salimi et al reported that over-scanning particularly in the inferior direction occurred in more than 95% of chest CT examinations [ 7 ]. Another study from Demircioglu et al reported a DICE score of 0.99 ± 0.1 for DL and radiologists’ annotations of scan range in routine chest CT [ 8 ].…”

Section: Discussionmentioning

confidence: 99%

“…Such monitoring can lead to implementation of mitigating steps to correct frequent errors in prescribed scan length. In fact, Salimi et al reported a 21% reduction in radiation dose for chest CT examinations with the use of a fully automated scan range definition from localizer or planning radiographs as compared to the manual prescription of scan length [ 7 ]. More work is, however, needed to develop and validate applications of such DL algorithms for identifying inappropriate scan coverage for other CT protocols, including those beyond the chest.…”

Section: Discussionmentioning

confidence: 99%

“…To assess the frequency and implication of over-scanning and under-scanning, a manual evaluation of images is time-consuming, subjective, inefficient, and expensive. Although there are prior reports [ 2 , 6 , 7 ] on the artificial intelligence (AI)-driven determination of scan coverage, our work differs from prior publications since it uses a physician-trained tool for determining scan coverage. Such tools can help physicians train AI models without any programming knowledge or experience to target areas of quality improvement and safety that the software or medical companies might not deem as critical revenue or profit-generating solutions.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Radiologist-Trained and -Tested (R2.2.4) Deep Learning Models for Identifying Anatomical Landmarks in Chest CT

et al. 2022

View full text Add to dashboard Cite

(1) Background: Optimal anatomic coverage is important for radiation-dose optimization. We trained and tested (R2.2.4) two (R3-2) deep learning (DL) algorithms on a machine vision tool library platform (Cognex Vision Pro Deep Learning software) to recognize anatomic landmarks and classify chest CT as those with optimum, under-scanned, or over-scanned scan length. (2) Methods: To test our hypothesis, we performed a study with 428 consecutive chest CT examinations (mean age 70 ± 14 years; male:female 190:238) performed at one of the four hospitals. CT examinations from two hospitals were used to train the DL classification algorithms to identify lung apices and bases. The developed algorithms were then tested on the data from the remaining two hospitals. For each CT, we recorded the scan lengths above and below the lung apices and bases. Model performance was assessed with receiver operating characteristics (ROC) analysis. (3) Results: The two DL models for lung apex and bases had high sensitivity, specificity, accuracy, and areas under the curve (AUC) for identifying under-scanning (100%, 99%, 99%, and 0.999 (95% CI 0.996–1.000)) and over-scanning (99%, 99%, 99%, and 0.998 (95%CI 0.992–1.000)). (4) Conclusions: Our DL models can accurately identify markers for missing anatomic coverage and over-scanning in chest CTs.

show abstract

Section: Discussionmentioning

confidence: 94%

Section: Discussionmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Radiologist-Trained and -Tested (R2.2.4) Deep Learning Models for Identifying Anatomical Landmarks in Chest CT

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Use of camera-assisted patient positioning/isocentre for CT scanning 15 , determining scan range 16 , and virtual MRI cockpits could mean that for high volume low-complexity or routine imaging with patients who have previously undergone imaging that the entire episode of care could be delivered remotely/autonomously in the not-too-distant future. Further examples of AI contribution to practice include the automation of image post processing and dose optimisation 17 .…”

Section: Radiography Practice Applicationsmentioning

confidence: 99%