A Regression Model for Predicting Shape Deformation after Breast Conserving Surgery

Hooshiar Zolfagharnasab, Sílvia Bessa, Sara Oliveira, Pedro Faria, João Teixeira, Jaime Cardoso, Hélder Oliveira
2018 Sensors  
Breast cancer treatments can have a negative impact on breast aesthetics, in case when surgery is intended to intersect tumor. For many years mastectomy was the only surgical option, but more recently breast conserving surgery (BCS) has been promoted as a liable alternative to treat cancer while preserving most part of the breast. However, there is still a significant number of BCS intervened patients who are unpleasant with the result of the treatment, which leads to self-image issues and
more » ... onal overloads. Surgeons recognize the value of a tool to predict the breast shape after BCS to facilitate surgeon/patient communication and allow more educated decisions; however, no such tool is available that is suited for clinical usage. These tools could serve as a way of visually sensing the aesthetic consequences of the treatment. In this research, it is intended to propose a methodology for predict the deformation after BCS by using machine learning techniques. Nonetheless, there is no appropriate dataset containing breast data before and after surgery in order to train a learning model. Therefore, an in-house semi-synthetic dataset is proposed to fulfill the requirement of this research. Using the proposed dataset, several learning methodologies were investigated, and promising outcomes are obtained. Sensors 2018, 18, 167 2 of 29 plus a margin of healthy tissues to eliminate cancerous cells. With better cosmesis results, BCS is nowadays the preferred alternative to mastectomy. Yet, a treatment plan is always tailored based on both medical and personal choices. Treatment options are conditioned by the biology of the tumor, the stage of breast cancer, the patient's health conditions and preferences [1] . Several studies have shown that the survival rate is almost the same for both mastectomy and BCS, with the benefit that the second imposes less deformation on breast and a more satisfactory aesthetic outcome can be achieved [4] . Still, despite the smaller deformation after BCS, it has been reported that up to 30% of patients are dissatisfied with their post-operative appearance [5] . Actually, the final aesthetic outcome can be affected by so many different variables, from different surgical practices and expertise, to some breast specific characteristics, such as volume and density, tumor size and location, hardening the prediction and patient/surgeon communication about surgical procedure results. Patients are usually involved in the decision process regarding their surgery, but most of the time surgeons lack the means to provide visual clues about the post-surgery results of different alternatives. Even though, this is an important step, regarding the acceptance of the final outcome, as also the contribution of breasts to the sense of femininity and beauty of most women. In fact, follow-up studies after breast cancer treatment show the harmful impact of poor aesthetic results on the psychosocial health of women, who describe loss of self-esteem [2], sexual impairment [6] and dislike towards their bodies after treatment [7] . On the other hand, physicians have been recognizing the value of support decision systems for planning BCS, to compare the outcome of different surgical options and facilitate surgeon/patient communication. The value of such systems is further supported by studies confirming that women are more willing to deal with the aesthetic results when they are included in the decision process [8]. These visual sensor tools could inform better the patient about the aesthetic consequences of the treatment and improve the feeling about all the process. The development of a planning/simulation tool for surgery demands the creation of three dimensional (3D) models of the breast that can be deformed in a realistic fashion, to reproduce known deformations imposed by surgery; however, the creation of such models is a challenging task due to the deformable characteristics of the breast, the lack of landmarks to define its shape and the complex nature of the deformations imposed by the surgery. To the best of our knowledge, there are currently no tools, other than surgical experience and clinical judgment [9] , to predict the impact of BCS on the shape and deformation of the treated breast [10] . In fact, the available solutions usually rely on generic models, are mainly targeted to plastic surgery (namely breast augmentation) and do not comprise complex deformations, such as the ones resulting from BCS. Moreover, they usually demand expensive and large equipment to scan patient's torso and require expertise to handle those scans [11] . Still, in literature, strategies to model breast deformations are abundant and designed to different applications: estimate pose transformation [12] [13] [14] , assist registration tasks among different radiological imaging modalities [15] [16] [17] , model breast deformation [18] [19] [20] , guide surgery [21, 22] , predict the healing process of the breast after tumor removal [23, 24] , among others. In particular, we highlight the work of Vavourakis et al. [24] , that proposed a 3D surgical simulator to predict a patient-specific outcome after BCS. This framework predicts the breast shape after surgery taking the wound healing process into account. The simulator relies on a coupled multiscale Finite Element (FE) numerical procedure to solve two mathematical models: a biochemical model for wound healing and angiogenesis, and a biomechanical model for soft tissues and pose estimation. The first considers both wound healing biochemical process and the formation of new blood vessels, while the second predicts the breast shape as function of the breast tissues mass density and the body force vector. The final shape of the breast is then predicted as an integration of both models. The aforementioned applications have in common the use of biomechanical models to predict deformations that, due to some inherent limitations, is not the most suitable approach to include in a tool designed to be used in the daily clinical practice. First, the computational process for most algorithms might take hours, days or even almost a week, depending on the complexity of the models. As a principle aim, to ease the patient/physician communications during their consultations,
doi:10.3390/s18010167 pmid:29315279 pmcid:PMC5795402 fatcat:iimb4lwuuvf57mvvjqqvppfqli