Background Among the long-term consequences of COVID-19, a disease caused by the SARS-CoV-2 virus, the worst is the development of pulmonary fibrosis which is associated with irreversible respiratory compromise if not identified and treated from the onset. It therefore appears essential to have tools that can predict its beginning already in the initial stages of the disease and, for this purpose, we are supported by texture analysis, a computerized analysis methodology that provides objective and quantitative assessments by analyzing radiological diagnostic imaging. A 2021 publication, which dealt with texture analysis with this aim, demonstrated that for some parameters extracted from the LIFEx software there is relevant statistical significance albeit with the limitations of a study population of only 14 subjects. Aim of the study Evaluate whether the parameters related to the onset of pulmonary fibrosis, identified through the HRTC texture analysis of the lung parenchyma of subjects in the acute phase of COVID-19 of the 2021 pilot study of the University of Padua, are confirmed as statistically significant also in a follow-up on a larger cohort. Furthermore, we want to determine whether a greater number allows to verify the association of other variables with the onset of post COVID-19 pulmonary fibrosis. Methods This is a retrospective observational study carried out in the period between March 2020 and May 2022 on COVID-19 patients treated at the Ospedali Riuniti Padova Sud. For these individuals CT images collected in the acute phase of SARS-CoV-2 pneumonia together with those acquired in the follow-up at least 6 months after onset were available at the U.O.C. Radiology. The acute phase CT investigations were analyzed using the LIFEx software, through which various texture parameters were calculated that were related to the presence or absence of pulmonary fibrosis in the follow-up CT. Statistical analysis was performed on the various extracted parameters using the MedCalc and R software to evaluate the significance of the variables under examination. Results A population of 140 patients (M / F = 83/57; mean age 66 ± 12 years, range 19-92 years) was tracked after hospital dismissal for a mean period of 324 ± 102 days (range 182-636 days). Two groups were created in relation to the presence or absence of pulmonary fibrosis at the last follow-up. The texture analysis conducted on the CT images was carried out on each individual lung for a total of 280 lungs analyzed through the LIFEx program which extracted 104 texture parameters; among these, statistical analysis identified 15 parameters with p <0.05. A statistically significant difference was therefore demonstrated between the two groups. For the INTENSITY-BASED first order statistics, the CoefficientOfVariation, QuartileCoefficientOfDispersion were significant, while for the INTENSITY-HISTOGRAM the Skewness, the 75thPercentile, the 90thPercentile, the InterquartileRange and the QuartileCoefficientOfDispersion were significant. Among the second-order statistics with adequate p there is for the GLCM the JointVariance, the DifferenceAverage, the Dissimilarity, the InverseDifference, while for the GLRLM the ShortRunsEmphasis; for NGTDM we identify Complexity as significant. Finally, the SmallZoneEmphasis and ZonePercentage were statistically valid for the GLSZM group. By far the features that contribute most to the difference between present and absent fibrosis are the QuartileCoefficientOfDispersion, the Skewness of the INTENSITY_HISTOGRAM and the JointVariance of GLCM. Conclusions The use of the texture analysis of the pulmonary parenchyma in the acute phase of COVID-19 can detect many parameters turned out useful in order to predict the onset of post-COVID-19 fibrosis
Presupposti: Tra le conseguenze a lungo termine del COVID-19, malattia causata dal virus SARS-CoV-2, la più infausta è lo sviluppo di fibrosi polmonare che si associa ad una compromissione respiratoria irreversibile se non identificata e trattata fin dagli esordi. Appare quindi fondamentale avere degli strumenti che possano prevederne l’insorgenza già nelle fasi iniziali di malattia e, a questo scopo, ci viene in supporto la texture analysis, metodologia di analisi computerizzata che fornisce valutazioni oggettive e quantitative analizzando imaging di diagnostica radiologica. Una pubblicazione del 2021, che ha trattato la texture analysis con questo obiettivo, ha dimostrato come per alcuni parametri estratti dal software LIFEx vi sia una rilevante significatività statistica sebbene con i limiti di una popolazione in studio di solo 14 soggetti. Scopo dello studio: Valutare se i parametri correlati all’insorgenza di fibrosi polmonare, identificati attraverso l’analisi texture di HRTC del parenchima polmonare di soggetti in fase acuta da COVID-19 dello studio pilota del 2021 dell’Università degli Studi di Padova, siano confermati come statisticamente significativi anche in un follow-up su di una coorte più ampia. Inoltre, si vuole determinare se una numerosità maggiore consenta di verificare l’associazione di altre variabili con l’insorgenza di fibrosi polmonare post COVID-19. Materiali e metodi: Si tratta di uno studio osservazionale retrospettivo svolto nel periodo compreso tra marzo 2020 e maggio 2022 su pazienti COVID-19 presi in carico presso gli Ospedali Riuniti Padova Sud. Di questi soggetti erano disponibili presso l’U.O.C. di Radiologia le immagini TC raccolte in fase acuta di polmonite da SARS-CoV-2 unitamente a quelle acquisite nel follow-up ad almeno 6 mesi dall’esordio. Le indagini TC in fase acuta sono state analizzate tramite il software LIFEx, attraverso il quale sono stati calcolati diversi parametri texture che sono stati messi in relazione alla presenza o all’assenza di fibrosi polmonare nelle TC del follow-up. Sui diversi parametri estratti è stata eseguita una analisi statistica tramite il software MedCalc e R al fine di valutarne la significatività di variabili in esame. Risultati: Una popolazione di 140 pazienti (M/F = 83/57; età media 66 ± 12 anni, intervallo 19-92 anni) è stata seguita dopo la dimissione ospedaliera per un periodo medio di 324 ± 102 giorni (intervallo 182-636 giorni). Sono stati creati due gruppi relativamente alla presenza ovvero assenza di fibrosi polmonare all’ultimo follow-up. L’analisi texture condotta sulle immagini TC è stata effettuata su ogni singolo polmone per un totale di 280 polmoni analizzati tramite il programma LIFEx che ha estratto 104 parametri texture; tra questi l’analisi statistica ne ha identificati 15 con p < 0,05. È stata quindi dimostrata una differenza statisticamente significativa tra i due gruppi. Per le statistiche del primo ordine INTENSITY-BASED son risultate significative la CoefficientOfVariation, QuartileCoefficientOfDispersion mentre per le INTENSITY-HISTOGRAM la Skewness, il 75thPercentile, il 90thPercentile, l’InterquartileRange e il QuartileCoefficientOfDispersion. Tra le statistiche di secondo ordine con p adeguato vi è per la GLCM la JointVariance, la DifferenceAverage, la Dissimilarity, la InverseDifference, mentre per la GLRLM lo ShortRunsEmphasis; per NGTDM identifichiamo come significativo il Complexity. Infine, per il gruppo GLSZM sono risultati statisticamente validi lo SmallZoneEmphasis, e ZonePercentage. Conclusioni: L’utilizzo dell’analisi texture del parenchima polmonare in fase acuta è capace di rilevare un numero consistente di parametri da permettere di individuareall’esordio i soggetti che manifesteranno fibrosi polmonare
UTILIZZO DELL'ANALISI TEXTURE DEL PARENCHIMA POLMONARE IN FASE ACUTA DA INFEZIONE DA COVID-19 PER PREVEDERE L'INSORGENZA DI FIBROSI
LANTIERI, ROBERTO
2021/2022
Abstract
Background Among the long-term consequences of COVID-19, a disease caused by the SARS-CoV-2 virus, the worst is the development of pulmonary fibrosis which is associated with irreversible respiratory compromise if not identified and treated from the onset. It therefore appears essential to have tools that can predict its beginning already in the initial stages of the disease and, for this purpose, we are supported by texture analysis, a computerized analysis methodology that provides objective and quantitative assessments by analyzing radiological diagnostic imaging. A 2021 publication, which dealt with texture analysis with this aim, demonstrated that for some parameters extracted from the LIFEx software there is relevant statistical significance albeit with the limitations of a study population of only 14 subjects. Aim of the study Evaluate whether the parameters related to the onset of pulmonary fibrosis, identified through the HRTC texture analysis of the lung parenchyma of subjects in the acute phase of COVID-19 of the 2021 pilot study of the University of Padua, are confirmed as statistically significant also in a follow-up on a larger cohort. Furthermore, we want to determine whether a greater number allows to verify the association of other variables with the onset of post COVID-19 pulmonary fibrosis. Methods This is a retrospective observational study carried out in the period between March 2020 and May 2022 on COVID-19 patients treated at the Ospedali Riuniti Padova Sud. For these individuals CT images collected in the acute phase of SARS-CoV-2 pneumonia together with those acquired in the follow-up at least 6 months after onset were available at the U.O.C. Radiology. The acute phase CT investigations were analyzed using the LIFEx software, through which various texture parameters were calculated that were related to the presence or absence of pulmonary fibrosis in the follow-up CT. Statistical analysis was performed on the various extracted parameters using the MedCalc and R software to evaluate the significance of the variables under examination. Results A population of 140 patients (M / F = 83/57; mean age 66 ± 12 years, range 19-92 years) was tracked after hospital dismissal for a mean period of 324 ± 102 days (range 182-636 days). Two groups were created in relation to the presence or absence of pulmonary fibrosis at the last follow-up. The texture analysis conducted on the CT images was carried out on each individual lung for a total of 280 lungs analyzed through the LIFEx program which extracted 104 texture parameters; among these, statistical analysis identified 15 parameters with p <0.05. A statistically significant difference was therefore demonstrated between the two groups. For the INTENSITY-BASED first order statistics, the CoefficientOfVariation, QuartileCoefficientOfDispersion were significant, while for the INTENSITY-HISTOGRAM the Skewness, the 75thPercentile, the 90thPercentile, the InterquartileRange and the QuartileCoefficientOfDispersion were significant. Among the second-order statistics with adequate p there is for the GLCM the JointVariance, the DifferenceAverage, the Dissimilarity, the InverseDifference, while for the GLRLM the ShortRunsEmphasis; for NGTDM we identify Complexity as significant. Finally, the SmallZoneEmphasis and ZonePercentage were statistically valid for the GLSZM group. By far the features that contribute most to the difference between present and absent fibrosis are the QuartileCoefficientOfDispersion, the Skewness of the INTENSITY_HISTOGRAM and the JointVariance of GLCM. Conclusions The use of the texture analysis of the pulmonary parenchyma in the acute phase of COVID-19 can detect many parameters turned out useful in order to predict the onset of post-COVID-19 fibrosisFile | Dimensione | Formato | |
---|---|---|---|
Lantieri_Roberto.pdf
accesso riservato
Dimensione
3.3 MB
Formato
Adobe PDF
|
3.3 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/30549