Ability of Radiomics Versus Humans in Predicting First-Pass Effect After Endovascular Treatment in the ESCAPE-NA1 Trial

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00159816%3A_____%2F23%3A00079663" target="_blank" >RIV/00159816:_____/23:00079663 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/00216224:14110/23:00133719

Výsledek na webu

<a href="https://www.ahajournals.org/doi/10.1161/SVIN.122.000525" target="_blank" >https://www.ahajournals.org/doi/10.1161/SVIN.122.000525</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1161/SVIN.122.000525" target="_blank" >10.1161/SVIN.122.000525</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Ability of Radiomics Versus Humans in Predicting First-Pass Effect After Endovascular Treatment in the ESCAPE-NA1 Trial

Popis výsledku v původním jazyce

BACKGROUND: First-pass effect (FPE), that is, achieving reperfusion with a single thrombectomy device pass, is associated with better clinical outcomes in patients with acute stroke. FPE is therefore increasingly used as a marker of device and procedural efficacy. We aimed to evaluate the ability of thrombus-based radiomics models to predict FPE in patients undergoing endovascular thrombectomy and compare performance with experts and nonradiomics thrombus characteristics. METHODS: Patients with thin-slice noncontrast computed tomography and computed tomography angiography from the ESCAPE-NA1 (Efficacy and Safety of Nerinetide for the Treatment of Acute Ischemic Stroke) trial were included. Thrombi were manually segmented on all images. Data were randomly split into a derivation set that included a training and a validation subset and an independent test set. Radiomics features were extracted from the derivation set. The machine learning models were compared with 3 expert stroke physicians in predicting FPE in the test set using area under the receiver operating characteristic curves. RESULTS: Thin-slice images of 554 patients were divided into a derivation set (training [n=388] and validation [n=55]) and a test set (n=111). A radiomics model using the combination of noncontrast computed tomography, computed tomography angiography, and noncontrast computed tomography-computed tomography angiography difference achieved the highest performance (area under the curve, 0.74 [95% CI, 0.64-0.84]) for prediction of FPE. This was higher than the mean area under the curve of the 3 experts (0.62 [95% CI, 0.53-0.71], P=0.01 for difference in area under the curves). The radiomics model also performed better than nonradiomics-based thrombus features such as volume and permeability measurements in predicting FPE (P&lt;0.05). Addition of device type did not improve the performance of the chosen radiomics model in predicting FPE. CONCLUSION: A radiomics-based machine learning model of thrombus characteristics from noncontrast computed tomography and computed tomography angiography performs better than experts and traditional nonradiomics imaging features in predicting FPE in patients with acute stroke treated with endovascular thrombectomy.

Název v anglickém jazyce

Ability of Radiomics Versus Humans in Predicting First-Pass Effect After Endovascular Treatment in the ESCAPE-NA1 Trial

Popis výsledku anglicky

BACKGROUND: First-pass effect (FPE), that is, achieving reperfusion with a single thrombectomy device pass, is associated with better clinical outcomes in patients with acute stroke. FPE is therefore increasingly used as a marker of device and procedural efficacy. We aimed to evaluate the ability of thrombus-based radiomics models to predict FPE in patients undergoing endovascular thrombectomy and compare performance with experts and nonradiomics thrombus characteristics. METHODS: Patients with thin-slice noncontrast computed tomography and computed tomography angiography from the ESCAPE-NA1 (Efficacy and Safety of Nerinetide for the Treatment of Acute Ischemic Stroke) trial were included. Thrombi were manually segmented on all images. Data were randomly split into a derivation set that included a training and a validation subset and an independent test set. Radiomics features were extracted from the derivation set. The machine learning models were compared with 3 expert stroke physicians in predicting FPE in the test set using area under the receiver operating characteristic curves. RESULTS: Thin-slice images of 554 patients were divided into a derivation set (training [n=388] and validation [n=55]) and a test set (n=111). A radiomics model using the combination of noncontrast computed tomography, computed tomography angiography, and noncontrast computed tomography-computed tomography angiography difference achieved the highest performance (area under the curve, 0.74 [95% CI, 0.64-0.84]) for prediction of FPE. This was higher than the mean area under the curve of the 3 experts (0.62 [95% CI, 0.53-0.71], P=0.01 for difference in area under the curves). The radiomics model also performed better than nonradiomics-based thrombus features such as volume and permeability measurements in predicting FPE (P&lt;0.05). Addition of device type did not improve the performance of the chosen radiomics model in predicting FPE. CONCLUSION: A radiomics-based machine learning model of thrombus characteristics from noncontrast computed tomography and computed tomography angiography performs better than experts and traditional nonradiomics imaging features in predicting FPE in patients with acute stroke treated with endovascular thrombectomy.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

30210 - Clinical neurology

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2023

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Stroke: Vascular and Interventional Neurology

ISSN

2694-5746

e-ISSN

—

Svazek periodika

3

Číslo periodika v rámci svazku

3

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

10

Strana od-do

—

Kód UT WoS článku

001162312700004

EID výsledku v databázi Scopus

—