Projeto Portugal 2020

Código de operação Código que identifica cada projeto efetuado com recurso a fundos europeus.

Data de conclusão

Sumário

Breast cancer is the most prevalent in women and the second most common overall, affecting 2.1M women each year [1,2]. In 2018, 627k deaths occurred due to breast cancer, representing 15% of all deaths from cancer in women[1]. To reduce mortality, early diagnosis and treatment are critical [3]. Currently, diagnosis of breast lesions is performed using palpation or medical imaging, such as mammography, ultrasound (US), or magnetic resonance (MR) [3]. After imaging screening, a biopsy is performed if a lesion is detected to characterize it as malign/benign [4]. Here, a needle is inserted to extract sample cells from the suspicious area using US guidance [5]. Also, when a non-palpable lesion must be removed, a US-guided preoperative localization procedure is performed after the biopsy, where a device is placed within the lesion to guide the surgeon intraoperatively on the lesion excision afterward [6]. Several limitations occur in the traditional approaches for breast lesions diagnosis and interventions. First is the dependency on US imaging. The usage of the US for the diagnosis has been growing over the last years [7]. Nevertheless, the US image acquisition and interpretation is still challenging and performed manually, thus, restricted to be used by specialized clinicians [7,8]. This difficulty, allied with the increasing number of breast cancer cases, tends to lead to a delay in the diagnosis and interventions, thus, a higher number of deaths. The second limitation is related to the breast tumor biopsies. Here, the physician needs to handle the US device at the same time that evaluates the image to infer the optimal puncture site and trajectory to reach the lesion [4]. Next, it needs to puncture the breast and track the needle trajectory to infer possible misalignments with its “mental map” of the optimal trajectory. The difficulty of this process is linked with the size and number of the lesions, as well as with the elastic properties of the breast that modify the anatomical position of the lesions when the US probe and needle presses the breast. Given the difficulty to perform a precise puncture, it is often the case that the surgeon needs to perform several needle insertion attempts to access the lesion target successfully. Moreover, the device used to aid lesion localization on excision surgery, which traditionally consists of an external wire, causes discomfort to the patient and can offer unintentional displaced or dislodged [11]. To minimise these issues, real-time image processing methods to accurately process intra-examination US images by delineating and classifying the lesions were proposed [9]. Moreover, medical robotics combined with US systems were already proposed to aid the traditional handheld procedure, namely, diagnosis and biopsies [10]. Regarding the lesion location for excision surgery, different types of markers were already proposed to replace the wire-based solution, such as radioactive or magnetic seeds, being the latter more adequate to avoid radioactivity [12,13]. Although interesting methodologies have been proposed to overcome problems related to specific steps of oncological breast treatment stages, an integrated and complete solution that could guide the physician throughout the entire breast treatment is missing. Particularly, the currently available solutions still do not allow to: i) aid the physician on handheld US image acquisition for optimal lesion identification and characterization - diagnosis stage; ii) combine the information of other image modalities (MR) for a correct assessment of the 3D lesion morphology - diagnosis, biopsy and perioperative stages; iii) combine the real-time image information and the prior knowledge of the lesion in augmented navigation environment - biopsy and lesion marking stages; iv) easily revisit the lesion for continuous monitoring of the lesion progression - long-term assessment of lesion highly suspicious for cancer; and, v) bring the valuable preoperative data (i.e. MRI and lesion region of excision-defined by the physician) into the intervention real-time lesion assessment - lesion excision surgery. To overcome these limitations and improve the efficacy and precision of breast cancer interventions we propose a novel framework - OncoNAVIGATOR. The operation plan starts with the study and characterization of the current workflow of breast oncological interventions (activity A1). Here, a literature review of works related to oncological interventions will be pursued (T1.2-T1.5). Next, system requirements, in compliance with European Union medical device regulations (MDR), will be defined (activity A2). First, breast interventions will be monitored to define functional requirements (T2.1). Then, the system requirements of each individual OncoNAVIGATOR module will be described, according to IEC 60601 and IEC 62304 (T2.2-T2.7). The general safety and performance requirements will be also studied to prevent risks according to ISO/TS 15066:2016 and IEC 80601-2-77, and usability and system validation tests are defined according to IEC 62336 (T2.8). At activity A3, the OncoNAVIGATOR modules will be developed. Thus, a US dataset of patients with different types of breast lesions will be constructed (T3.1). From that, in T3.2, AI methods to extract and interpret information from US and MR data will be developed. Next, a wire/radiation-free magnetic localization system will be developed to allow precise, accurate, and real-time breast tumor mapping (T3.3). Then, we propose to combine the information from the real-time US and the magnetic tracker with a collaborative medical robot to precisely locate the breast lesion while guaranteeing a safe and correct needle insertion of the market in the lesion. For it, a US/Magnetic-guided collaborative robotic module for robot-assisted breast intervention will be investigated (T3.4) and an intelligent kinematic model for safe robot usage to increase its manipulability will be explored (T3.5). In T3.6 a natural user interface to guide the clinician through the treatment will be developed. Finally, the developed modules will be combined in activity A4 with the construction of OncoNAVIGATOR prototypes. At activity A5, usability tests of the developed prototypes (T5.1), in-vitro studies with phantoms (T5.2), and animal experiments (T5.3) will be pursued to demonstrate the added-value of OncoNAVIGATOR. In activity A6 it is envisioned to develop a business model (T6.1) and to protect the intellectual property of the project (T6.2). During the project, the dissemination of results achieved will be made with the creation of promotion material (T7.1), participation in international conferences (T7.2), and publication on international journals (T7.3), culminating with the organization of a seminar to present all the project results (T7.4).