Calculation of Silicone Breast Implant Volumes Using Breast Magnetic Resonance Imaging

Methods 

This study was Institutional Review Board–approved and complied with the Health Insurance Portability and Accountability Act. Breast MRI scans were obtained for medical necessity before implant revision surgery or for follow-up on 20 patients (for a total of 39 breasts) from one plastic surgeon (RR). In 18 patients, the indication for a MRI scan was evaluation of implant integrity. One patient had a MRI scan for follow-up on a nodule. In only one patient—who had Poland syndrome treated with multiple implants in another country—was the MRI done clinically to establish implant size before revision. Fifteen patients (27 breasts) had known implant volumes. One radiologist (NF) performed the CAD volume computations on all patients. Actual implant volumes were either blinded (for volumes based on preoperative notes) or unknown at the time of calculation.

The CAD program will apply color to the white pixels (silicone) on axial T2 images. In some cases, it is necessary to adjust the colorization threshold to highlight the implant only, as shown in Figure 3, Figure 4. Using the volume of interest tool, a box is created around each silicone implant. Figure 5 shows an axial view with the implant centered inside the box. The operator can quickly scroll through the images in all three planes to verify that: (1) the implant is completely enclosed by the region of interest, and that (2) the colorization threshold is adjusted to include as much of the implant as possible while causing little or no colorization of areas around the implant. The CAD program calculates the percentage of colorized pixels within the volume of interest (reported in cm3).

Clinical volumes to compare with MRI calculations were determined for 27 breasts from either documented preoperative notes before the original implantation (n = 17) or after removal by volume markings on the implants (n = 10). Deviations of MRI volumes from known clinical volumes were calculated using the following equation:

Results 

Clinical and calculated silicone gel–filled implant volume data are presented in the Table. The average deviation of the 27 known implant volumes from those calculated by MRI DynaCAD analysis was only +0.82% (SD, 3.95%). Four of 27 (14.8%) MRI calculations matched the clinical volumes exactly. The widest volume deviation ranges for all implants were +11.4% to −4.44%. A variety of implant types were found by previous records and 11 of 27 implants were confirmed to be ruptured at surgery, with no correlation of implant types or rupture with range variations. The average length of time required to obtain volume calculations was five minutes.

Table. Clinical comparison to calculated volumes

	Patient	Side	Known volume, cc (preoperative)	Known volume, cc (postoperative)	MRI CAD volume (cc)	Difference (vs. preoperative)	Difference (vs. postoperative)	Implant type*	Rupture
	A	R	270	—	265	−1.85 %	—	Round	No
		L	270	—	270	0.00 %	—	Round	No
	B	R	450	—	435	−3.33 %	—	Round	No
		L	450	—	430	−4.44 %	—	Round	No
	C	R	220	—	231	5.00 %	—	Double lumen	Yes
		L	220	—	228	3.64 %	—	Double lumen	Yes
	D	R	720	—	730	1.39 %	—	Style 153	No
		L	720	—	693	−3.75 %	—	Style 153	No
	E	L	225	—	217	−3.56 %	—	Style 410	No
	F	R	230	—	242	5.22 %	—	Double lumen	Yes
		L	230	—	225	−2.17 %	—	Double lumen	No
	G	R	450	—	480	6.67 %	—	Anatomic	Yes
		L	450	—	459	2.00 %	—	Anatomic	Yes
	H	R	270	—	267	−1.11 %	—	Misti	No
		L	270	—	270	0.00 %	—	Misti	No
	I	R	280	—	296	5.71 %	—	Special fill	Yes
		L	280	—	290	3.57 %	—	Special fill	No
	J	R	—	300	291	—	−3.00 %	Round†	Yes
		L	—	300	292	—	−2.67 %	Round†	Yes
	K	R	—	200	207	—	3.50%	Round†	Yes
		L	—	200	196	—	−2.00 %	Round†	Yes
	L	R	—	220	220	—	0.00 %	Round	No
		L	—	240	242	—	0.83 %	Round	No
	M	R	—	500	557	—	11.4 %	Round	No
	N	R	—	230	230	—	0.00 %	Round	No
		L	—	230	222	—	−3.48 %	Round	No
	O	L	—	200	209	—	4.50 %	Round†	Yes

CAD, Computer-assisted detection; L, left; MRI, magnetic resonance imaging; R, right.

* Implant type taken from operative notes or implant stickers. † Presumed round by appearance at surgery.

Of the five breast implants in which the calculated volume was >5% over the actual volume, two were double lumen, one “anatomic,” one “special fill,” and one round. Therefore, four of five high outliers were implant types other than standard round. No physical or MRI finding were found that could predict inaccurate calculations.

Discussion 

The purpose of our study was to develop a methodology for calculating implant volumes and to compare them with actual volumes known either from a documented preoperative note or postoperative volume via implant markings.

Volume estimates can be calculated manually from MRI images by measuring the implant area for a series of slices of known spacing.3, 4, 5 In 1995, Mineyev et al4 studied MRI determination of breast implant volume and, at that time, concluded that the method was unreliable. More recently, Akbas et al6 showed the successful and more rapid use of MRI to estimate breast implant volumes. They employed the Cavalieri principle, where a series of random, parallel, two-dimensional slices of known spacing are used to estimate volume. Even so, using the CAD features of MRI systems holds the possibility of performing volume calculations more rapidly. We found that DynaCAD (Invivo, Birmingham, MI) calculations could be performed in about five minutes.

CAD systems were initially developed to help radiologists quickly analyze kinetic data from dynamic contrast-enhanced breast MRI scans performed for breast cancer detection. These studies typically acquire pre- and postcontrast images over a period of several minutes. Subtraction images show areas of enhancement as bright signal intensity. The CAD program uses various colors to highlight and quantify areas of contrast enhancement over time. CAD was not designed to evaluate silicone implants, but can be used to do so. Breast MRI for implant evaluation does not require intravenous contrast enhancement because silicone is characterized by bright signal intensity on T2 images, as shown in Figure 1, Figure 2.

CAD will colorize bright signal intensity and render three-dimensional reconstruction images. With this in mind, we wondered whether CAD could be used to calculate the volume of silicone implants. We are aware of no methods that could provide this information quickly and accurately.

We used the DynaCAD workstation (a commercially available CAD system) to colorize the silicone on T2 scans selectively. A volume of interest square or rectangle was placed around the colorized silicone implant in all three planes. Slices must be contiguous. The program then quickly calculated the implant volume based on the percentage of colorized “volumetric” pixels. The technique could also be used for saline implants because saline will stand out as white on T2 and could be colorized. However, MRI scans are not performed to determine saline implant integrity—the main indication for the MRI scans in this study—because saline rupture is clinically obvious. Whether it is practical to have a patient get an expensive MRI solely for volume determination would have to be carefully individualized.

The results of our study indicate that the average deviation of our MRI scan–derived volumes from known volumes is quite low. This suggests that silicone gel–filled breast implant volumes can be accurately measured before implant revision surgery, thereby allowing the correct sized implants to be available at time of surgery.

Conclusions 

Calculation of silicone breast implant volume using breast MRI scans and a commercially available CAD system appears to be sufficiently accurate that it may have clinical benefit in planning revision implant surgery. To our knowledge, this is the first report describing a method in which silicone breast implant volumes can be readily and reliably ascertained before implant revision surgery.