Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening: systematic review of test accuracy studies

Julia Geppert; Asra Asgharzadeh; Anna Brown; Chris Stinton; Emma J Helm; Surangi   Jayakody; Daniel  Todkill; Daniel Gallacher; Hesam Ghiasvand; Mubarak Patel; Peter Auguste; Alexander  Tsertsvadze; Yen-Fu Chen; Amy Grove; Bethany Shinkins; Aileen Clarke; Sian Taylor Phillips

doi:10.1136/thorax-2024-221662

Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening: systematic review of test accuracy studies

Julia Geppert
, Asra Asgharzadeh
, Anna Brown
, Chris Stinton
, Emma J Helm
, Surangi Jayakody
, Daniel Todkill
, Daniel Gallacher
, Hesam Ghiasvand
, Mubarak Patel
, Peter Auguste
, Alexander Tsertsvadze
, Yen-Fu Chen
, Amy Grove
, Bethany Shinkins
, Aileen Clarke
, Sian Taylor Phillips

Research output: Contribution to journal › Article › peer-review

28 Citations (Scopus)

31 Downloads (Pure)

Abstract

Objectives: To examine the accuracy and impact of artificial intelligence (AI) software assistance in lung cancer screening using computed tomography (CT).
Methods: A systematic review of CE-marked, AI-based software for automated detection and analysis of nodules in CT lung cancer screening was conducted. Multiple databases including Medline, Embase and Cochrane CENTRAL were searched from 2012 to March 2023. Primary research reporting test accuracy or impact on reading time or clinical management was included. QUADAS2/QUADAS-C were used to assess risk of bias. We undertook narrative synthesis.
Results: Eleven studies evaluating six different AI-based programs and reporting on 19,770 patients were eligible. All were at high risk of bias with multiple applicability concerns. Compared to unaided reading, AI-assisted reading was faster and generally improved sensitivity (+5% to +20% for detecting/categorising actionable nodules; +3% to +15% for detecting/categorising malignant nodules), with lower specificity (-7% to -3% for detecting/categorising actionable nodules; -8% to -2% for detecting/categorising malignant nodules). AI assistance tended to increase the proportion of nodules allocated to higher risk categories. Assuming 0.5% cancer prevalence, these results would translate into additional 150 to 750 cancers detected per million participants but lead to an additional 59,700 to 79,600 participants without cancer receiving unnecessary CT surveillance.
Conclusions: AI assistance in lung cancer screening may improve sensitivity but increases the number of false positive results and unnecessary surveillance. Future research needs to increase the specificity of AI-assisted reading and minimise risk of bias and applicability concerns through improved study design.

Original language	English
Pages (from-to)	1040-1049
Number of pages	10
Journal	Thorax
Volume	79
Issue number	11
DOIs	https://doi.org/10.1136/thorax-2024-221662
Publication status	Published - 25 Sept 2024

Bibliographical note

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made.

Funding

This review was funded by the UK National Institute for Health and Care Research (NIHR) Evidence Synthesis Programme (NIHR135325). The funder had no role in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the article for publication. DT and AG are partly funded by the NIHR West Midlands Applied Research Collaboration. STP is funded by the NIHR through a research professorship (NIHR302434). AG is supported by a NIHR Fellowship (NIHR300060).

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Access to Document

10.1136/thorax-2024-221662Licence: CC BY

Ghiasvand2024VORFinal published version, 580 KBLicence: CC BY

Cite this

Geppert, J., Asgharzadeh, A., Brown, A., Stinton, C., Helm, E. J., Jayakody, S., Todkill, D., Gallacher, D., Ghiasvand, H., Patel, M., Auguste, P., Tsertsvadze, A., Chen, Y.-F., Grove, A., Shinkins, B., Clarke, A., & Taylor Phillips, S. (2024). Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening: systematic review of test accuracy studies. Thorax, 79(11), 1040-1049. https://doi.org/10.1136/thorax-2024-221662

Geppert, J, Asgharzadeh, A, Brown, A, Stinton, C, Helm, EJ, Jayakody, S, Todkill, D, Gallacher, D, Ghiasvand, H, Patel, M, Auguste, P, Tsertsvadze, A, Chen, Y-F, Grove, A, Shinkins, B, Clarke, A & Taylor Phillips, S 2024, 'Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening: systematic review of test accuracy studies', Thorax, vol. 79, no. 11, pp. 1040-1049. https://doi.org/10.1136/thorax-2024-221662

@article{51420567a15d4b85b9152ae3aef15098,

title = "Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening: systematic review of test accuracy studies",

abstract = "Objectives: To examine the accuracy and impact of artificial intelligence (AI) software assistance in lung cancer screening using computed tomography (CT). Methods: A systematic review of CE-marked, AI-based software for automated detection and analysis of nodules in CT lung cancer screening was conducted. Multiple databases including Medline, Embase and Cochrane CENTRAL were searched from 2012 to March 2023. Primary research reporting test accuracy or impact on reading time or clinical management was included. QUADAS2/QUADAS-C were used to assess risk of bias. We undertook narrative synthesis. Results: Eleven studies evaluating six different AI-based programs and reporting on 19,770 patients were eligible. All were at high risk of bias with multiple applicability concerns. Compared to unaided reading, AI-assisted reading was faster and generally improved sensitivity (+5\% to +20\% for detecting/categorising actionable nodules; +3\% to +15\% for detecting/categorising malignant nodules), with lower specificity (-7\% to -3\% for detecting/categorising actionable nodules; -8\% to -2\% for detecting/categorising malignant nodules). AI assistance tended to increase the proportion of nodules allocated to higher risk categories. Assuming 0.5\% cancer prevalence, these results would translate into additional 150 to 750 cancers detected per million participants but lead to an additional 59,700 to 79,600 participants without cancer receiving unnecessary CT surveillance. Conclusions: AI assistance in lung cancer screening may improve sensitivity but increases the number of false positive results and unnecessary surveillance. Future research needs to increase the specificity of AI-assisted reading and minimise risk of bias and applicability concerns through improved study design. ",

author = "Julia Geppert and Asra Asgharzadeh and Anna Brown and Chris Stinton and Helm, \{Emma J\} and Surangi Jayakody and Daniel Todkill and Daniel Gallacher and Hesam Ghiasvand and Mubarak Patel and Peter Auguste and Alexander Tsertsvadze and Yen-Fu Chen and Amy Grove and Bethany Shinkins and Aileen Clarke and \{Taylor Phillips\}, Sian",

note = "This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made.",

year = "2024",

month = sep,

day = "25",

doi = "10.1136/thorax-2024-221662",

language = "English",

volume = "79",

pages = "1040--1049",

journal = "Thorax",

issn = "0040-6376",

publisher = "BMJ Publishing Group",

number = "11",

}

TY - JOUR

T1 - Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening

T2 - systematic review of test accuracy studies

AU - Geppert, Julia

AU - Asgharzadeh, Asra

AU - Brown, Anna

AU - Stinton, Chris

AU - Helm, Emma J

AU - Jayakody, Surangi

AU - Todkill, Daniel

AU - Gallacher, Daniel

AU - Ghiasvand, Hesam

AU - Patel, Mubarak

AU - Auguste, Peter

AU - Tsertsvadze, Alexander

AU - Chen, Yen-Fu

AU - Grove, Amy

AU - Shinkins, Bethany

AU - Clarke, Aileen

AU - Taylor Phillips, Sian

N1 - This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made.

PY - 2024/9/25

Y1 - 2024/9/25

N2 - Objectives: To examine the accuracy and impact of artificial intelligence (AI) software assistance in lung cancer screening using computed tomography (CT). Methods: A systematic review of CE-marked, AI-based software for automated detection and analysis of nodules in CT lung cancer screening was conducted. Multiple databases including Medline, Embase and Cochrane CENTRAL were searched from 2012 to March 2023. Primary research reporting test accuracy or impact on reading time or clinical management was included. QUADAS2/QUADAS-C were used to assess risk of bias. We undertook narrative synthesis. Results: Eleven studies evaluating six different AI-based programs and reporting on 19,770 patients were eligible. All were at high risk of bias with multiple applicability concerns. Compared to unaided reading, AI-assisted reading was faster and generally improved sensitivity (+5% to +20% for detecting/categorising actionable nodules; +3% to +15% for detecting/categorising malignant nodules), with lower specificity (-7% to -3% for detecting/categorising actionable nodules; -8% to -2% for detecting/categorising malignant nodules). AI assistance tended to increase the proportion of nodules allocated to higher risk categories. Assuming 0.5% cancer prevalence, these results would translate into additional 150 to 750 cancers detected per million participants but lead to an additional 59,700 to 79,600 participants without cancer receiving unnecessary CT surveillance. Conclusions: AI assistance in lung cancer screening may improve sensitivity but increases the number of false positive results and unnecessary surveillance. Future research needs to increase the specificity of AI-assisted reading and minimise risk of bias and applicability concerns through improved study design.

AB - Objectives: To examine the accuracy and impact of artificial intelligence (AI) software assistance in lung cancer screening using computed tomography (CT). Methods: A systematic review of CE-marked, AI-based software for automated detection and analysis of nodules in CT lung cancer screening was conducted. Multiple databases including Medline, Embase and Cochrane CENTRAL were searched from 2012 to March 2023. Primary research reporting test accuracy or impact on reading time or clinical management was included. QUADAS2/QUADAS-C were used to assess risk of bias. We undertook narrative synthesis. Results: Eleven studies evaluating six different AI-based programs and reporting on 19,770 patients were eligible. All were at high risk of bias with multiple applicability concerns. Compared to unaided reading, AI-assisted reading was faster and generally improved sensitivity (+5% to +20% for detecting/categorising actionable nodules; +3% to +15% for detecting/categorising malignant nodules), with lower specificity (-7% to -3% for detecting/categorising actionable nodules; -8% to -2% for detecting/categorising malignant nodules). AI assistance tended to increase the proportion of nodules allocated to higher risk categories. Assuming 0.5% cancer prevalence, these results would translate into additional 150 to 750 cancers detected per million participants but lead to an additional 59,700 to 79,600 participants without cancer receiving unnecessary CT surveillance. Conclusions: AI assistance in lung cancer screening may improve sensitivity but increases the number of false positive results and unnecessary surveillance. Future research needs to increase the specificity of AI-assisted reading and minimise risk of bias and applicability concerns through improved study design.

UR - https://www.scopus.com/pages/publications/85205452471

U2 - 10.1136/thorax-2024-221662

DO - 10.1136/thorax-2024-221662

M3 - Article

SN - 0040-6376

VL - 79

SP - 1040

EP - 1049

JO - Thorax

JF - Thorax

IS - 11

ER -

Software using artificial intelligence for nodule and cancer detection in CT lung cancer screening: systematic review of test accuracy studies

Abstract

Bibliographical note

Funding

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this