Mammographic Estimates of Tumour to Breast Volume to Improve Oncoplastic Decision-Making
Association of Breast Surgery ePoster Library. Soosainathan A. 05/15/17; 166300; P101
Ms. Arany Soosainathan
REGULAR CONTENT
Login now to access Regular content available to all registered users.
Abstract
Rate & Comment (0)
Introduction
Decisions regarding breast conserving surgery (BCS), oncoplastic surgery (OPS), or mastectomy involve balancing many clinical and radiological factors to achieve complete resection and optimise cosmesis. Regarding BCS, cosmetic satisfaction decreases with volume of resection, yet surgeons have no objective methods to estimate breast tumour to volume ratios (TVR). Algorithmic predictions of TVR using mammographic imaging may provide additional information critical for oncoplastic decision-making.
Methods
Conventional CC and MLO mammographic images were obtained from n=100 patients. Machine learning algorithms (Partial Least Squares Regression and Primary Component Regression) were applied, calculating breast volume, tumour volume, and TVR. Based on these estimates, comparisons were made between predicted and real operative decisions.
Results
In n=57 patients, the surgical plan was concordant with algorithmic predictions. Of discordant decisions, n=5 patients had less extensive procedures than predicted [n=2 downstaged with neoadjuvant chemotherapy, n=2 required re-operative interventions for positive margins, n=1 refused surgery]. In n=38 patients, more extensive procedures were performed than predicted [n=10 patient choice, n=6 tumours found to be radio-occult/multifocal on further imaging, n=3 tumours in 'unfavourable' sites for BCS, n=5 had OPS rather than BCS for better cosmesis, n=1 had risk-reducing surgery for BRCA]. In n=13 cases, discordant decisions were unclear.
Conclusions
Software offering TVR estimates provides further objective and useful information for surgical decision-making. While decisions regarding BCS versus mastectomy are complex and multifactorial, information on the percentage predicted resection volume may alter decision-making to minimise re-operative surgeries and optimise aesthetic outcomes.
Decisions regarding breast conserving surgery (BCS), oncoplastic surgery (OPS), or mastectomy involve balancing many clinical and radiological factors to achieve complete resection and optimise cosmesis. Regarding BCS, cosmetic satisfaction decreases with volume of resection, yet surgeons have no objective methods to estimate breast tumour to volume ratios (TVR). Algorithmic predictions of TVR using mammographic imaging may provide additional information critical for oncoplastic decision-making.
Methods
Conventional CC and MLO mammographic images were obtained from n=100 patients. Machine learning algorithms (Partial Least Squares Regression and Primary Component Regression) were applied, calculating breast volume, tumour volume, and TVR. Based on these estimates, comparisons were made between predicted and real operative decisions.
Results
In n=57 patients, the surgical plan was concordant with algorithmic predictions. Of discordant decisions, n=5 patients had less extensive procedures than predicted [n=2 downstaged with neoadjuvant chemotherapy, n=2 required re-operative interventions for positive margins, n=1 refused surgery]. In n=38 patients, more extensive procedures were performed than predicted [n=10 patient choice, n=6 tumours found to be radio-occult/multifocal on further imaging, n=3 tumours in 'unfavourable' sites for BCS, n=5 had OPS rather than BCS for better cosmesis, n=1 had risk-reducing surgery for BRCA]. In n=13 cases, discordant decisions were unclear.
Conclusions
Software offering TVR estimates provides further objective and useful information for surgical decision-making. While decisions regarding BCS versus mastectomy are complex and multifactorial, information on the percentage predicted resection volume may alter decision-making to minimise re-operative surgeries and optimise aesthetic outcomes.
Introduction
Decisions regarding breast conserving surgery (BCS), oncoplastic surgery (OPS), or mastectomy involve balancing many clinical and radiological factors to achieve complete resection and optimise cosmesis. Regarding BCS, cosmetic satisfaction decreases with volume of resection, yet surgeons have no objective methods to estimate breast tumour to volume ratios (TVR). Algorithmic predictions of TVR using mammographic imaging may provide additional information critical for oncoplastic decision-making.
Methods
Conventional CC and MLO mammographic images were obtained from n=100 patients. Machine learning algorithms (Partial Least Squares Regression and Primary Component Regression) were applied, calculating breast volume, tumour volume, and TVR. Based on these estimates, comparisons were made between predicted and real operative decisions.
Results
In n=57 patients, the surgical plan was concordant with algorithmic predictions. Of discordant decisions, n=5 patients had less extensive procedures than predicted [n=2 downstaged with neoadjuvant chemotherapy, n=2 required re-operative interventions for positive margins, n=1 refused surgery]. In n=38 patients, more extensive procedures were performed than predicted [n=10 patient choice, n=6 tumours found to be radio-occult/multifocal on further imaging, n=3 tumours in 'unfavourable' sites for BCS, n=5 had OPS rather than BCS for better cosmesis, n=1 had risk-reducing surgery for BRCA]. In n=13 cases, discordant decisions were unclear.
Conclusions
Software offering TVR estimates provides further objective and useful information for surgical decision-making. While decisions regarding BCS versus mastectomy are complex and multifactorial, information on the percentage predicted resection volume may alter decision-making to minimise re-operative surgeries and optimise aesthetic outcomes.
Decisions regarding breast conserving surgery (BCS), oncoplastic surgery (OPS), or mastectomy involve balancing many clinical and radiological factors to achieve complete resection and optimise cosmesis. Regarding BCS, cosmetic satisfaction decreases with volume of resection, yet surgeons have no objective methods to estimate breast tumour to volume ratios (TVR). Algorithmic predictions of TVR using mammographic imaging may provide additional information critical for oncoplastic decision-making.
Methods
Conventional CC and MLO mammographic images were obtained from n=100 patients. Machine learning algorithms (Partial Least Squares Regression and Primary Component Regression) were applied, calculating breast volume, tumour volume, and TVR. Based on these estimates, comparisons were made between predicted and real operative decisions.
Results
In n=57 patients, the surgical plan was concordant with algorithmic predictions. Of discordant decisions, n=5 patients had less extensive procedures than predicted [n=2 downstaged with neoadjuvant chemotherapy, n=2 required re-operative interventions for positive margins, n=1 refused surgery]. In n=38 patients, more extensive procedures were performed than predicted [n=10 patient choice, n=6 tumours found to be radio-occult/multifocal on further imaging, n=3 tumours in 'unfavourable' sites for BCS, n=5 had OPS rather than BCS for better cosmesis, n=1 had risk-reducing surgery for BRCA]. In n=13 cases, discordant decisions were unclear.
Conclusions
Software offering TVR estimates provides further objective and useful information for surgical decision-making. While decisions regarding BCS versus mastectomy are complex and multifactorial, information on the percentage predicted resection volume may alter decision-making to minimise re-operative surgeries and optimise aesthetic outcomes.
Code of conduct/disclaimer available in General Terms & Conditions
{{ help_message }}
{{filter}}