polito.it
Politecnico di Torino (logo)

Instance segmentation for food-based pathologies

Mattia Rizzo

Instance segmentation for food-based pathologies.

Rel. Tatiana Tommasi. Politecnico di Torino, Corso di laurea magistrale in Mechatronic Engineering (Ingegneria Meccatronica), 2021

Abstract:

The tracking of nutrients in aliments is a key aspect to consider when dealing with several pathologies. A clear example is type-1 diabetes: people diagnosed with this disease need to administrate insulin injections on daily basis. The estimation of the prandial dose is complex, time consuming and depends on several factors. Studies demonstrated that even well-trained patients, tent to over/underestimate CHO by 27% on average on every meal. Moreover, especially for early diagnosed people with little to none training, it could be hard to keep track of the CHO contained in each food, especially when they go eating outside. Type 1 diabetics are not the only patients that suffer the disadvantages and the complexity of their treatments. There are indeed other pathologies such as obesity that require a careful control of the food, in particular calories. One interesting case is represented by irritable bowel syndrome. This is a pathology which is strictly related to patients’ diet because people with this syndrome cannot digest some type of foods called FODMAPs. FODMAPs are a group of short-chain carbohydrates that are either poorly absorbed in the small intestine or are completely indigestible. Their poor absorption in some people triggers IBS symptoms. Currently there are many commercially available apps that allow you to have a list of foods with several characteristics for each ingredient. A good example is the Fodmap app developed by Monash University: it has a big database of foods and dishes with some nutrients, and it indicates which foods are fodmap free and which one is fodmap heavy, so it should not be eaten. The main problem with this type of solution is that it is inconvenient to use, since it implies a manual research of each ingredient or each dish. With this research we aim to facilitate the use and the identification of fodmap food and other properties using computer vision. Thank to this, we allow people to have a direct feedback of what they are eating, allowing also to track the calories and the properties of the food based on the ingredients and the quantity.

Relatori: Tatiana Tommasi
Anno accademico: 2020/21
Tipo di pubblicazione: Elettronica
Numero di pagine: 78
Informazioni aggiuntive: Tesi secretata. Fulltext non presente
Soggetti:
Corso di laurea: Corso di laurea magistrale in Mechatronic Engineering (Ingegneria Meccatronica)
Classe di laurea: Nuovo ordinamento > Laurea magistrale > LM-25 - INGEGNERIA DELL'AUTOMAZIONE
Ente in cotutela: Brigham and Women's Hospital (STATI UNITI D'AMERICA)
Aziende collaboratrici: Harvard Medical School
URI: http://webthesis.biblio.polito.it/id/eprint/22347
Modifica (riservato agli operatori) Modifica (riservato agli operatori)