NIRS and biophysical analyses on cassava cooking properties report. High-Throughput Phenotyping Protocols (HTPP), WP3

Davrieux Fabrice, Belalcazar Martinez John Eiver, Zhang Xiaofei, Tran Thierry. 2021. NIRS and biophysical analyses on cassava cooking properties report. High-Throughput Phenotyping Protocols (HTPP), WP3. Saint-Pierre : RTBfoods Project-CIRAD, 18 p.

Résumé : This scientific report concerns data analysis of two matrices of measured data on fresh cassava 1) physico-chemical data and 2) spectral data. The data were collected on fresh cassava in CIAT, Colombia. The analyses were performed using 87 cassava genotypes: 38 genotypes were analysed once, 37 analysed twice and 12 analysed 4 times. The total number of analyses is 160. The samples were analysed for their cooking properties (cooking time in boiling water), texture parameters (gradient, max force, distance at max force, area, linear distance and end force/ max force), dry matter content and water absorption capacity during cooking. The same genotypes were analysed using near infrared spectroscopy. The absorption spectra were performed on ground samples of fresh roots using a FOSS 2500 spectrometer. The average dry matter is 40,4%, which is constant over months (age of the root). The mean value of cooking time is 33 min, ranging from 10 to 57 min. The wide distribution of cooking time allows to group the samples into 3 classes: less or equal to 25 min; higher than 25 min and lower or equal to 40 min; and higher than 40 min. Water absorption values at 10, 20 , 30 min are highly correlated (r >= ?). There is a non-linear relation between water absorption at 20 min and optimum cooking time: Genotypes with longer cooking time absorb less water at 20 min than “good cooking” genotypes. The values of gradient range between 225 and 2247 kg/mm with an average of 1179 kg/mm, which follows a normal distribution. Gradient is highly correlated to physical values related to Max force, Area and Linear distance. Gradient is also correlated to optimum cooking time (r = 0,735). The correlation between gradient and water absorption at 20 min of boiling is significant with r = –0,601, the highest correlation is at 40 min of boiling (r = -0,792). The relation between gradient and water absorption at 20 min of boiling is non linear (second order), genotypes with high gradient values absorb less water at 20 min of boiling than genotypes with low gradient values that showed low optimum cooking time. Different multivariate approaches were investigated to associate spectral data and physico-chemical parameters. The direct calibrations of physico-chemical parameters were not performant. Classification according to three cooking time classes was tested using different algorithms. Whatever were the pretreatments used (SNV, SNVD, first or second derivative…) and whatever the classification approach (K Nearest Neighbors, Support Vector Machine, Naive Bayesian Classifier, Random Forest, Classification Regression Trees…), the predictions of a validation set for the 3 cooking time classes failed. The best classification method was obtained by doing a prediction of the scores of the discriminant axes calculated on six physico-chemical variables (DM, WA10, WA20, OCT, Gradient and distance at max force). The best classification was obtained for two cooking time classes: ≤ 30 min and > 30 min. The classification successful rate, for a validation set, was 80%. The performances of the classification method which mix laboratory values and spectra values indicate that spectra contain relevant information related to cooking properties, and confirm that deep learning approaches may help for better and faster classification.

Mots-clés libres : Cassava, Cooking properties, Near Infrared Spectroscopy, Water absorption, Classification, PLS regression

Agences de financement hors UE : Bill and Melinda Gates Foundation, Centre de Coopération Internationale en Recherche Agronomique pour le Développement, Centro Internacional de Agricultura Tropical, Institut National de Recherche pour l'Agriculture, l'Alimentation et l'Environnement, James Hutton Institute

Projets sur financement : (FRA) Breeding RTB Products for End User Preferences

Auteurs et affiliations

• Davrieux Fabrice, CIRAD-PERSYST-UMR Qualisud (REU)
• Belalcazar Martinez John Eiver, CIRAD-PERSYST-UMR Qualisud (FRA)
• Zhang Xiaofei, CIAT (COL)
• Tran Thierry, CIRAD-PERSYST-UMR Qualisud (COL) ORCID: 0000-0002-9557-3340

Source : Cirad-Agritrop (https://agritrop.cirad.fr/603867/)

Métriques

[ Page générée et mise en cache le 2023-04-21 ]