Deep learning model utilizing clinical data alone outperforms image-based model for hernia recurrence following abdominal wall reconstruction with long-term follow up.
Abdominal wall reconstruction
Artificial intelligence
Deep learning
Hernia recurrence
Ventral hernia
Journal
Surgical endoscopy
ISSN: 1432-2218
Titre abrégé: Surg Endosc
Pays: Germany
ID NLM: 8806653
Informations de publication
Date de publication:
11 Jun 2024
11 Jun 2024
Historique:
received:
13
04
2023
accepted:
02
06
2024
medline:
12
6
2024
pubmed:
12
6
2024
entrez:
11
6
2024
Statut:
aheadofprint
Résumé
Deep learning models (DLMs) using preoperative computed tomography (CT) imaging have shown promise in predicting outcomes following abdominal wall reconstruction (AWR), including component separation, wound complications, and pulmonary failure. This study aimed to apply these methods in predicting hernia recurrence and to evaluate if incorporating additional clinical data would improve the DLM's predictive ability. Patients were identified from a prospectively maintained single-institution database. Those who underwent AWR with available preoperative CTs were included, and those with < 18 months of follow up were excluded. Patients were separated into a training (80%) set and a testing (20%) set. A DLM was trained on the images only, and another DLM was trained on demographics only: age, sex, BMI, diabetes, and history of tobacco use. A mixed-value DLM incorporated data from both. The DLMs were evaluated by the area under the curve (AUC) in predicting recurrence. The models evaluated data from 190 AWR patients with a 14.7% recurrence rate after an average follow up of more than 7 years (mean ± SD: 86 ± 39 months; median [Q1, Q3]: 85.4 [56.1, 113.1]). Patients had a mean age of 57.5 ± 12.3 years and were majority (65.8%) female with a BMI of 34.2 ± 7.9 kg/m The clinical-only DLM outperformed both the image-only DLM and the mixed-value DLM in predicting recurrence. While all three models were poorly predictive of recurrence, the clinical-only DLM was the most predictive. These findings may indicate that imaging characteristics are not as useful for predicting recurrence as they have been for other AWR outcomes. Further research should focus on understanding the imaging characteristics that are identified by these DLMs and expanding the demographic information incorporated in the clinical-only DLM to further enhance the predictive ability of this model.
Sections du résumé
BACKGROUND
BACKGROUND
Deep learning models (DLMs) using preoperative computed tomography (CT) imaging have shown promise in predicting outcomes following abdominal wall reconstruction (AWR), including component separation, wound complications, and pulmonary failure. This study aimed to apply these methods in predicting hernia recurrence and to evaluate if incorporating additional clinical data would improve the DLM's predictive ability.
METHODS
METHODS
Patients were identified from a prospectively maintained single-institution database. Those who underwent AWR with available preoperative CTs were included, and those with < 18 months of follow up were excluded. Patients were separated into a training (80%) set and a testing (20%) set. A DLM was trained on the images only, and another DLM was trained on demographics only: age, sex, BMI, diabetes, and history of tobacco use. A mixed-value DLM incorporated data from both. The DLMs were evaluated by the area under the curve (AUC) in predicting recurrence.
RESULTS
RESULTS
The models evaluated data from 190 AWR patients with a 14.7% recurrence rate after an average follow up of more than 7 years (mean ± SD: 86 ± 39 months; median [Q1, Q3]: 85.4 [56.1, 113.1]). Patients had a mean age of 57.5 ± 12.3 years and were majority (65.8%) female with a BMI of 34.2 ± 7.9 kg/m
CONCLUSIONS
CONCLUSIONS
The clinical-only DLM outperformed both the image-only DLM and the mixed-value DLM in predicting recurrence. While all three models were poorly predictive of recurrence, the clinical-only DLM was the most predictive. These findings may indicate that imaging characteristics are not as useful for predicting recurrence as they have been for other AWR outcomes. Further research should focus on understanding the imaging characteristics that are identified by these DLMs and expanding the demographic information incorporated in the clinical-only DLM to further enhance the predictive ability of this model.
Identifiants
pubmed: 38862826
doi: 10.1007/s00464-024-10980-y
pii: 10.1007/s00464-024-10980-y
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Informations de copyright
© 2024. The Author(s).
Références
Ciomperlik H, Dhanani NH, Cassata N, Mohr C, Bernardi K, Holihan JL, Lyons N, Olavarria O, Ko TC, Liang MK (2021) Patient quality of life before and after ventral hernia repair. Surgery 169:1158–1163. https://doi.org/10.1016/j.surg.2020.11.003
doi: 10.1016/j.surg.2020.11.003
pubmed: 33317902
Cox TC, Huntington CR, Blair LJ, Prasad T, Lincourt AE, Heniford BT, Augenstein VA (2016) Predictive modeling for chronic pain after ventral hernia repair. Am J Surg 212:501–510. https://doi.org/10.1016/j.amjsurg.2016.02.021
doi: 10.1016/j.amjsurg.2016.02.021
pubmed: 27443426
Shao JM, Deerenberg EB, Elhage SA, Prasad T, Davis BR, Kercher KW, Colavita PD, Augenstein VA, Heniford BT (2021) Recurrent incisional hernia repairs at a tertiary hernia center: are outcomes really inferior to initial repairs? Surgery 169:580–585. https://doi.org/10.1016/j.surg.2020.10.009
doi: 10.1016/j.surg.2020.10.009
pubmed: 33248712
Heniford BT, Ross SW, Wormer BA, Walters AL, Lincourt AE, Colavita PD, Kercher KW, Augenstein VA (2020) Preperitoneal ventral hernia repair. Ann Surg 271:364–374. https://doi.org/10.1097/SLA.0000000000002966
doi: 10.1097/SLA.0000000000002966
pubmed: 30080725
Flum DR, Horvath K, Koepsell T (2003) Have outcomes of incisional hernia repair improved with time? Ann Surg 237:129–135. https://doi.org/10.1097/00000658-200301000-00018
doi: 10.1097/00000658-200301000-00018
pubmed: 12496540
pmcid: 1513979
Holihan JL, Alawadi Z, Martindale RG, Roth SJ, Wray CJ, Ko TC, Kao LS, Liang MK (2015) Adverse events after ventral hernia repair: the vicious cycle of complications. J Am Coll Surg 221:478–485. https://doi.org/10.1016/j.jamcollsurg.2015.04.026
doi: 10.1016/j.jamcollsurg.2015.04.026
pubmed: 26206646
Poulose BK, Shelton J, Phillips S, Moore D, Nealon W, Penson D, Beck W, Holzman MD (2012) Epidemiology and cost of ventral hernia repair: making the case for hernia research. Hernia 16:179–183. https://doi.org/10.1007/s10029-011-0879-9
doi: 10.1007/s10029-011-0879-9
pubmed: 21904861
Köckerling F, Koch A, Lorenz R, Schug-Pass C, Stechemesser B, Reinpold W (2015) How long do we need to follow-up our hernia patients to find the real recurrence rate? Front Surg. https://doi.org/10.3389/fsurg.2015.00024
doi: 10.3389/fsurg.2015.00024
pubmed: 26779487
pmcid: 4657277
Liang MK, Holihan JL, Itani K, Alawadi ZM, Gonzalez JRF, Askenasy EP, Ballecer C, Sen CH, Goldblatt MI, Greenberg JA, Harvin JA, Keith JN, Martindale RG, Orenstein S, Richmond B, Roth JS, Szotek P, Towfigh S, Tsuda S, Vaziri K, Berger DH (2017) Ventral hernia management. Ann Surg 265:80–89. https://doi.org/10.1097/SLA.0000000000001701
doi: 10.1097/SLA.0000000000001701
pubmed: 28009730
Rosen MJ, Krpata DM, Petro CC, Carbonell A, Warren J, Poulose BK, Costanzo A, Tu C, Blatnik J, Prabhu AS (2022) Biologic vs synthetic mesh for single-stage repair of contaminated ventral hernias. JAMA Surg 157:293. https://doi.org/10.1001/jamasurg.2021.6902
doi: 10.1001/jamasurg.2021.6902
pubmed: 35044431
pmcid: 8771431
Tryliskyy Y, Wong CS, Demykhova I, Tyselskyi V, Kebkalo A, Poylin V, Pournaras DJ (2022) Fascial defect closure versus bridged repair in laparoscopic ventral hernia mesh repair: a systematic review and meta-analysis of randomized controlled trials. Hernia 26:1473–1481. https://doi.org/10.1007/s10029-021-02533-2
doi: 10.1007/s10029-021-02533-2
pubmed: 34748092
Olavarria OA, Bernardi K, Dhanani NH, Lyons NB, Harvin JA, Millas SG, Ko TC, Kao LS, Liang MK (2021) Synthetic versus biologic mesh for complex open ventral hernia repair: a pilot randomized controlled trial. Surg Infect (Larchmt) 22:496–503. https://doi.org/10.1089/sur.2020.166
doi: 10.1089/sur.2020.166
pubmed: 33259771
Demetrashvili Z, Pipia I, Loladze D, Metreveli T, Ekaladze E, Kenchadze G, Khutsishvili K (2017) Open retromuscular mesh repair versus onlay technique of incisional hernia: a randomized controlled trial. Int J Surg 37:65–70. https://doi.org/10.1016/j.ijsu.2016.12.008
doi: 10.1016/j.ijsu.2016.12.008
pubmed: 27940291
Parker SG, Mallett S, Quinn L, Wood CPJ, Boulton RW, Jamshaid S, Erotocritou M, Gowda S, Collier W, Plumb AAO, Windsor ACJ, Archer L, Halligan S (2021) Identifying predictors of ventral hernia recurrence: systematic review and meta-analysis. BJS Open. https://doi.org/10.1093/bjsopen/zraa071
doi: 10.1093/bjsopen/zraa071
pubmed: 34131707
pmcid: 8205854
Bernardi K, Adrales GL, Hope WW, Keith J, Kuhlens H, Martindale RG, Melin AA, Orenstein SB, Roth JS, Shah SK, Tsuda S, Liang MK (2018) Abdominal wall reconstruction risk stratification tools: a systematic review of the literature. Plast Reconstr Surg 142:9S-20S. https://doi.org/10.1097/PRS.0000000000004833
doi: 10.1097/PRS.0000000000004833
pubmed: 30138260
Loftus TJ, Tighe PJ, Filiberto AC, Efron PA, Brakenridge SC, Mohr AM, Rashidi P, Upchurch GR, Bihorac A (2020) Artificial intelligence and surgical decision-making. JAMA Surg 155:148. https://doi.org/10.1001/jamasurg.2019.4917
doi: 10.1001/jamasurg.2019.4917
pubmed: 31825465
pmcid: 7286802
Elhage SA, Deerenberg EB, Ayuso SA, Murphy KJ, Shao JM, Kercher KW, Smart NJ, Fischer JP, Augenstein VA, Colavita PD, Heniford BT (2021) Development and validation of image-based deep learning models to predict surgical complexity and complications in abdominal wall reconstruction. JAMA Surg 156:933. https://doi.org/10.1001/jamasurg.2021.3012
doi: 10.1001/jamasurg.2021.3012
pubmed: 34232255
pmcid: 8264757
Ayuso SA, Elhage SA, Zhang Y, Aladegbami BG, Gersin KS, Fischer JP, Augenstein VA, Colavita PD, Heniford BT (2023) Predicting rare outcomes in abdominal wall reconstruction using image-based deep learning models. Surgery 173:748–755. https://doi.org/10.1016/j.surg.2022.06.048
doi: 10.1016/j.surg.2022.06.048
pubmed: 36229252
Hassan AM, Lu S-C, Asaad M, Liu J, Offodile AC, Sidey-Gibbons C, Butler CE (2022) Novel machine learning approach for the prediction of hernia recurrence, surgical complication, and 30-day readmission after abdominal wall reconstruction. J Am Coll Surg 234:918–927. https://doi.org/10.1097/XCS.0000000000000141
doi: 10.1097/XCS.0000000000000141
pubmed: 35426406
Hope WW, Abdul W, Winters R (2023) Abdominal Wall Reconstruction. In: StatPearls. StatPearls, Treasure Island
Katzen MM, Kercher KW, Sacco JM, Ku D, Scarola GT, Davis BR, Colavita PD, Augenstein VA, Heniford BT (2023) Open preperitoneal ventral hernia repair: prospective observational study of quality improvement outcomes over 18 years and 1,842 patients. Surgery 173:739–747. https://doi.org/10.1016/j.surg.2022.07.042
doi: 10.1016/j.surg.2022.07.042
pubmed: 36280505
Shao JM, Ayuso SA, Deerenberg EB, Elhage SA, Prasad T, Colavita PD, Augenstein VA, Heniford BT (2022) Biologic mesh is non-inferior to synthetic mesh in CDC class 1 & 2 open abdominal wall reconstruction. Am J Surg 223:375–379. https://doi.org/10.1016/j.amjsurg.2021.05.019
doi: 10.1016/j.amjsurg.2021.05.019
pubmed: 34140156
Katzen M, Ayuso SA, Sacco J, Ku D, Scarola GT, Kercher KW, Colavita PD, Augenstein VA, Heniford BT (2023) Outcomes of biologic versus synthetic mesh in CDC class 3 and 4 open abdominal wall reconstruction. Surg Endosc 37:3073–3083. https://doi.org/10.1007/s00464-022-09486-2
doi: 10.1007/s00464-022-09486-2
pubmed: 35925400
Carter JV, Pan J, Rai SN, Galandiuk S (2016) ROC-ing along: evaluation and interpretation of receiver operating characteristic curves. Surgery 159:1638–1645. https://doi.org/10.1016/j.surg.2015.12.029
doi: 10.1016/j.surg.2015.12.029
pubmed: 26962006
Blair LJ, Ross SW, Huntington CR, Watkins JD, Prasad T, Lincourt AE, Augenstein VA, Heniford BT (2015) Computed tomographic measurements predict component separation in ventral hernia repair. J Surg Res 199:420–427. https://doi.org/10.1016/j.jss.2015.06.033
doi: 10.1016/j.jss.2015.06.033
pubmed: 26169031
Schlosser KA, Maloney SR, Prasad T, Colavita PD, Augenstein VA, Heniford BT (2020) Three-dimensional hernia analysis: the impact of size on surgical outcomes. Surg Endosc 34:1795–1801. https://doi.org/10.1007/s00464-019-06931-7
doi: 10.1007/s00464-019-06931-7
pubmed: 31236720
Schlosser KA, Maloney SR, Prasad T, Colavita PD, Augenstein VA, Heniford BT (2020) Too big to breathe: predictors of respiratory failure and insufficiency after open ventral hernia repair. Surg Endosc 34:4131–4139. https://doi.org/10.1007/s00464-019-07181-3
doi: 10.1007/s00464-019-07181-3
pubmed: 31637601
Love MW, Warren JA, Davis S, Ewing JA, Hall AM, Cobb WS, Carbonell AM (2021) Computed tomography imaging in ventral hernia repair: can we predict the need for myofascial release? Hernia 25:471–477. https://doi.org/10.1007/s10029-020-02181-y
doi: 10.1007/s10029-020-02181-y
pubmed: 32277369
Al-Mansour MR, Wu J, Gagnon G, Knee A, Romanelli JR, Seymour NE (2021) Linear versus volumetric CT analysis in predicting tension-free fascial closure in abdominal wall reconstruction. Hernia 25:91–98. https://doi.org/10.1007/s10029-020-02349-6
doi: 10.1007/s10029-020-02349-6
pubmed: 33389214
Choi RY, Coyner AS, Kalpathy-Cramer J, Chiang MF, Campbell JP (2020) Introduction to machine learning, neural networks, and deep learning. Transl Vis Sci Technol 9:14. https://doi.org/10.1167/tvst.9.2.14
doi: 10.1167/tvst.9.2.14
pubmed: 32704420
pmcid: 7347027
Ballem N, Parikh R, Berber E, Siperstein A (2008) Laparoscopic versus open ventral hernia repairs: 5 year recurrence rates. Surg Endosc 22:1935–1940. https://doi.org/10.1007/s00464-008-9981-1
doi: 10.1007/s00464-008-9981-1
pubmed: 18528613
Muysoms FE, Miserez M, Berrevoet F, Campanelli G, Champault GG, Chelala E, Dietz UA, Eker HH, El Nakadi I, Hauters P, Hidalgo Pascual M, Hoeferlin A, Klinge U, Montgomery A, Simmermacher RKJ, Simons MP, Śmietański M, Sommeling C, Tollens T, Vierendeels T, Kingsnorth A (2009) Classification of primary and incisional abdominal wall hernias. Hernia 13:407–414. https://doi.org/10.1007/s10029-009-0518-x
doi: 10.1007/s10029-009-0518-x
pubmed: 19495920
pmcid: 2719726
Holihan JL, Li LT, Askenasy EP, Greenberg JA, Keith JN, Martindale RG, Roth JS, Liang MK (2016) Analysis of model development strategies: predicting ventral hernia recurrence. J Surg Res 206:159–167. https://doi.org/10.1016/j.jss.2016.07.042
doi: 10.1016/j.jss.2016.07.042
pubmed: 27916356