LEON-ROQUE Prediction, fermentation, color ANN TALANTA

Prediction of fermentation index of cacao beans (Theobroma cacao L.) based on color measurement and artificial neural networks

• Noemí Leon-Roque^a, 
• Mohamed Abderrahim^b, 
• Luis Nuñez-Alejos^c, 
• Silvia M. Arribas^d, 
• Luis Condezo-Hoyos^b

• ^a Universidad Nacional Pedro Ruiz Gallo, Facultad de Ingeniería Química e Industrias Alimentarias, Departamento de Ingeniería en Industrias Alimentarias, Lambayeque, Peru
• ^b Universidad Carlos III de Madrid, Departamento de Sistemas y Automática, Leganés, Madrid, Spain
• ^c Programa Nacional de Alimentación-QaliWarma, Jaén, Cajamarca, Peru
• ^d Universidad Autónoma de Madrid, Facultad de Medicina, Departamento de Fisiología, Madrid, Spain

Received 13 May 2016, Revised 31 July 2016, Accepted 4 August 2016, Available online 5 August 2016

Highlights

•

A quantitative color-based model to predict fermentation in cacao beans was developed.

•

Artificial neural networks (ANN) were used to develop the model.

•

RGB values of beans surface and extracts were included in ANN model as predictors.

•

Model was validated using Bland-Altman plot and Passing-Bablok regression analysis.

ABSTRACT

Several procedures are currently used to assess fermentation index (FI) of cocoa beans (Theobroma cacao L.) for quality control. However, all of them present several drawbacks. The aim of the present work was to develop and validate a simple image based quantitative procedure, using color measurement and artificial neural network (ANNs). 

ANN models based on color measurements were tested to predict fermentation index (FI) of fermented cacao beans. The RGB values were measured from surface and center region of fermented beans in images obtained by camera and desktop scanner. The FI was defined as the ratio of total free amino acids in fermented versus non-fermented samples. 

The ANN model that included RGB color measurement of fermented cacao surface and R/G ratio in cacao bean of alkaline extracts was able to predict FI with no statistical difference compared with the experimental values. Performance of ANN model was evaluated by coefficient of determination, Bland-Altman plot and Passing-Bablok regression analyses. 

Moreover, in fermented beans, total sugar content and titratable acidity showed a similar pattern to the total free amino acid predicted through the color based ANN model. The results of the present work demonstrate that the proposed ANN model can be adopted as a low-cost and in situ procedure to predict FI in fermented cacao beans through apps developed for mobile device.

SOURCE