A tissue preparation to characterize uterine fibroid tissue properties for thermal therapies

Abstract

Purpose

Treating uterine fibroids with less invasive therapies such as magnetic resonance‐guided focused ultrasound (MRgFUS) is an attractive alternative to surgery. Treatment planning can improve MRgFUS procedures and reduce treatment times, but the tissue properties that currently inform treatment planning tools are not adequate. This study aims to develop an ex vivo uterine fibroid model that can emulate the in vivo environment allowing for characterization of the uterus and fibroid MR, acoustic, and thermal tissue properties while maintaining viability for the necessary postsurgical histopathological assessments.

Methods

Women undergoing a hysterectomy due to fibroid‐related symptoms were invited to undergo a preoperative pelvic MRI and to permit postoperative testing of their uterine specimen. Patients that declined or could not be scheduled for a pre‐operative MRI were still able to allow post‐operative testing of their excised tissue. Following surgical removal of the uterus, nonmorcellated tissues were reperfused with a Krebs‐Henseleit buffer solution. An MR‐compatible perfusion system was designed to maintain tissue viability inside the MR suite during scanning. MR imaging protocols utilized preoperatively were repeated on whole sample, reperfused ex vivo uterus specimens. Thermal properties including thermal diffusivity and thermal conductivity of the uterus and fibroids were determined using an invasive needle sensor device in 50% of the specimens. Acoustic property measurements (density, speed of sound and attenuation) were obtained for approximately 20% of the tissue samples using both through‐transmission and radiation force balance techniques. Differences between fibroid and uterus and in vivo and ex vivo measurements were evaluated with a two‐tailed Student t test.

Results

Fourteen patients participated in the study and measurements were obtained from 22 unique fibroids. Of the 16 fibroids available for preoperative MRI testing, 69% demonstrated classic hypo‐intensity relative to the myometrium, with the remainder presenting with iso‐ (25%) or hyper‐intensity (6%). While thermal diffusivity was not significantly different between fibroid and myometrium tissues (0.217 ± 0.047 and 0.204 ± 0.039 mm²/s, respectively), the acoustic attenuation in fibroid tissue was significantly higher than myometrium (0.092 ± 0.021 and 0.052 ± 0.023 Np/cm/MHz, respectively). When comparing in vivo with ex vivo MRI T1 and T2 measurements in fibroids and myometrium tissue, the only difference was found in the fibroid T2 property (P < 0.05). Finally, the developed perfusion protocol successfully maintained tissue viability in ex vivo tissues as evaluated through histological analysis.

Conclusions

This study developed an MR‐compatible extracorporeal perfusion technique that effectively maintains tissue viability, allowing for the direct measurement of patient‐specific MR, thermal, and acoustic property values for both fibroid and myometrium tissues. These measured tissue property values will enable further development and validation of treatment planning models that can be utilized during MRgFUS uterine fibroid treatments.

1. Introduction

Uterine fibroids, also known as leiomyomas, are the most common tumor in reproductive age females, occurring in nearly three‐quarters of women by age fifty.¹ While benign, fibroids causing debilitating symptoms such as excessive menstrual bleeding, pelvic pain, increased urinary frequency, and infertility are seen in approximately one‐quarter of cases.^{2, 3, 4} While the gold standard treatment for uterine fibroids is hysterectomy,⁵ there is growing interest in alternative procedures that preserve the uterus and avoid invasive procedures in general.^{2, 6}

Magnetic resonance‐guided focused ultrasound (MRgFUS) is an FDA‐approved treatment for uterine fibroids in which ultrasound is focused deep within the tumor under image guidance, quickly increasing the local tissue temperature and inducing necrosis, with magnetic resonance imaging (MRI) used for diagnosis, planning, monitoring, and treatment assessment.⁷ MRgFUS has been shown to significantly reduce uterine fibroid bulk and associated symptoms.^{8, 9, 10} The completely noninvasive nature of MRgFUS has advantages of eliminating surgery and potential infections and complications, maintaining the possibility of child‐bearing,^{10, 11} and significantly reducing recovery times.¹² One challenge slowing clinical acceptance of uterine fibroid MRgFUS treatments is the long procedure time.¹³ Pretreatment and on‐line modeling has the potential to reduce procedure times and improve efficacy through targeted ablations,¹⁴ optimized heating patterns^{15, 16, 17} and model‐predictive control.¹⁸ However, such models will only be as useful as the accuracy of the tissue properties that inform them. Variability in tissue physiological measurements can lead to important variations in the temperature elevation dynamic and significantly affect treatment predictions.¹⁹

Unfortunately, property values in the literature specific to myometrium and uterine fibroids are scarce. The IT'IS Foundation, which aims to provide a comprehensive up‐to‐date collection of tissue properties from the literature, lists one or two values for uterus thermal properties and a single value for each acoustic and MR property.²⁰ Some of those values were based exclusively on water percentage formulas instead of direct measurements, and the database includes no tissue property values specific to uterine fibroids. Acoustic properties of myometrium and uterine fibroids have been measured in other in vivo ^{21, 22} and ex vivo ²³ studies.

Additionally, patient‐specific properties would be preferable for treatment modeling purposes. However, accurate and precise measurements of uterus and fibroid thermal and acoustic properties in vivo are difficult due to issues including difficulty of in situ measurements, additional treatment time and tissue heterogeneity. There are often limitations in the clinical temperature measurements from which those properties may be derived including lack of temperature data during tissue cooling period, limited field of view 2D imaging and lower quality images due to reduced signal‐to‐noise ratio available in MRgFUS treatments.^{24, 25} With a more complete understanding of uterus and fibroid properties, it may be possible to better correlate acoustic and thermal properties to MR properties (that are easily quantified in vivo) for reliable and effective patient‐specific treatment planning.

This study takes the first steps towards that goal of quantifying uterus and fibroid properties by developing an ex vivo uterine fibroid model that can accurately replicate the in vivo environment and maintain tissue viability following hysterectomy. The model will provide opportunities to accurately characterize uterus and fibroid MR, acoustic, and thermal tissue properties without compromising the histopathological assessments required following surgery; initial tissue property data from the ex vivo uterine fibroid model are presented herein.

2. Materials and methods

2.A. IRB, recruitment and timeline overview

Women undergoing a hysterectomy due to fibroid‐related symptoms at the University of Utah Hospital and Huntsman Cancer Institute were invited to (a) undergo a preoperative pelvic MRI and (b) permit postoperative testing of their uterine specimen. Potential candidates for the study were identified by screening surgery schedules and initially approached about the study by their primary care physician. All study procedures were performed with patient consent and approval from the University of Utah Institutional Review Board (#00066145). A total of fourteen patients participated in the study and measurements were obtained from 22 unique fibroids.

Immediately following surgical removal of the uterus, each specimen was transferred from the operating room to the imaging center (<10 min transit) for ex vivo testing. The ex vivo testing included reperfusion (for whole samples only), ex vivo MRI, thermal property determination, gross pathology postsurgical processing, and acoustic property determination. Details for each of these steps, including study constraints, are outlined below.

2.B. In vivo MRI

The preoperative MRI was performed in the week prior to surgery on Siemens Tim 3T scanners (MAGNETOM Trio or Prisma^fit, Erlangen, Germany) with protocols based on the institution's standard pelvic MRI exam. The patient was positioned feet‐first‐supine on a 32‐channel spine array coil in the MR bore with an 18‐channel body matrix coil placed over the pelvis. Following localization, T2‐weighted images along the long and short axes of the uterus were used to ascertain the uterine size and anatomy, as well as fibroid size, location and signal intensity. Quantitative two‐dimensional maps for T2 and T1 were acquired in primary fibroids and the myometrium using a clinical T1/T2 mapping package (MyoMaps, Siemens Healthcare, Erlangen, Germany).^{26, 27} T1‐weighted dynamic contrast enhanced (DCE) imaging was also performed to characterize blood flow to the uterus and fibroids (0.2 ml/kg MultiHance, Bracco Diagnostics Inc., Singen, Germany). The imaging protocol, shown in Table 1 with scans applied in multiple imaging planes, was completed in 45–60 min. Imaging data were made available to physicians to assist in surgery planning.

Table 1.

MR imaging protocol used for both in vivo and ex vivo imaging of the uterus. Matrix size was adjusted for ex vivo imaging

MR sequence	TR (ms)	TE (ms)	Matrix size	Resolution (mm)	Slices	Bandwidth (Hz/pixel)	Other
T2w, Turbo Spin Echo	3700	94	200 × 200	0.8 × 0.8 × 4.0	20	170	Turbo factor = 15
T2 map, True FISP27	3000	1.66	360 × 322	2.5 × 1.9 × 6.0	1	500	T2 prep pulse
T1 map, True FISP26	2000	1.25	360 × 322	2.0 × 1.4 × 6.0	1	700	Nonselective inversion recovery with T1 prep
T1w DCE, TWIST	3.14	1.28	260 × 260	1.8 × 1.4 × 4.0	32	500	5.6 s acquisition

2.C. Cannulation and tissue preparation

Depending on the size of the uterus and fibroids, both whole and morcellated samples were obtained. While properties of morcellated tissue specimens were determined immediately, uterus specimens that were removed as whole specimen samples were reperfused as described below before measuring tissue properties. Depending on arterial size and access, catheters between 14 and 24 gauge were selected for cannulating each of the bilateral uterine arteries. Catheters were sutured to the tissue to mitigate back flow and prevent unintentional removal. Following cannulation, specimens were flushed with a Modified Krebs‐Henseleit Buffer,^{28, 29} an isotonic solution that is fortified with glucose, heparin, and insulin. Once the buffer perfusate passively exiting the uterine veins appeared clear (~30 min), specimens were submerged in a recirculating bath of the perfusate maintained at a constant temperature of 37°C.

2.D. Ex vivo model

Figure 1 presents both a schematic and photograph of the MR‐compatible experimental setup designed to maintain tissue viability. While other studies have developed systems for reperfusion of the uterus,^{28, 29} this work required modifications to operate in the MR environment.

Figure 1.

(a) Schematic and (b) photograph of experimental setup for maintaining tissue viability in the MR environment. Inset in (b) is placed outside the 10 gauss line. A ① temperature‐controlled water bath heats the ② perfusate to 37°C, which is then driven by a ③ peristaltic pump in ④ two flow lines to the MRI bore. Because the lines' temperature drops en route to the MRI bore, a ⑤ MRI‐compatible water bath (with temperature‐controlled water recirculated via a separate ⑥ pump) reheats the perfusate before it enters the cannulated uterine arteries of the ⑦uterus. Another ⑧ pump delivers heated perfusate to the cylinder in which the uterus is immersed. Excess perfusate is collected and recirculated with a fourth ⑨ pump. Water below the uterus acoustically couples a ⑩ MRgFUS system to the uterus [Color figure can be viewed at wileyonlinelibrary.com]

A temperature‐controlled water bath maintained at 37°C heated the perfusate, while a separate peristaltic pump — set at a baseline flow rate of ~15 ml/min — approximated physiological flow conditions,³⁰ feeding both cannulated uterine arteries. A second pump delivered perfusate to the cylinder in which the specimen was submerged at a fixed temperature. A third pump exchanged heated water between the temperature‐controlled bath and a MRI‐compatible bath inside the MRI bore. Because temperatures in the low flow rate lines feeding the uterine arteries decreased en route to the MRI bore, this MRI‐compatible water bath was used to reheat the perfusate to physiological temperatures before it reached the specimen. An MRgFUS system was acoustically coupled to the cylinder holding the uterus with a water bath, to acquire data for a separate study. Finally, a fourth pump collected perfusate to be rewarmed and recirculated. A filter was utilized to prevent debris from creating occlusions in specimen vasculature, and each flow line was fitted with a bubble trap to prevent gas occlusions.

Temperature (mercury‐in‐glass thermometer, Messgerate‐Werk Lauda, Lauda, Germany), pressure (mercury manometer attached to bubble traps, Riester, Jungingen, Germany), and pH (pen type pH probe, Etekcity pH‐2011, Anaheim, CA, USA) of the perfusate were monitored periodically during experiments. Adjustments to perfusate temperature were made by changing the set point of the temperature‐controlled water bath, pressure adjustments could be made by altering perfusate flow rate, and perfusate pH could be controlled by bubbling the perfusate with medical oxygen and/or carbon dioxide.

2.E. Ex vivo MRI

In these experiments, the cylinder holding the submerged tissue specimen was fitted with a custom single‐channel radiofrequency loop coil to improve local MR signal quality. The preoperative MR imaging protocols described in Table 1 were modified for field of view and repeated on the ex vivo specimen. In‐line ports just proximal to the specimen were used for introducing contrast agent for DCE imaging. Water‐filled bags were used to position the specimen in the cylinder.

2.F. Thermal property determination

Following the ex vivo MRI, a commercially available device (KD2 Pro Thermal Properties Analyzer, Decagon Devices, Pullman, WA, USA) was used to measure thermal diffusivity and thermal conductivity of the uterus and fibroids.³¹ Because the two heating and measurement probes on the device measure 3 cm in length and require 1 cm of tissue on all sides of the probe, quantifying thermal properties was only possible in larger fibroids approaching this size. The manufacturer reports a measurement accuracy of ±10% for both thermal conductivity and diffusivity in materials with a conductivity greater than 0.1 W/m/°C.

2.G. Histological analysis

Following thermal property measurements, specimens were delivered to the gross pathology lab for standard postsurgical processing and analysis. Hematoxylin and Eosin (H&E) stained sections were reviewed by the study pathologist (EAJ) after the original histopathologic diagnoses had been verified. In each case, the presence or absence of fixation artifact and/or any other morphologic alterations in the tissue potentially attributable to the reperfusion procedure was evaluated.

2.H. MRI analysis

Quantitative T1 and T2 MR slice orientations were prescribed to maximize the cross section of the fibroids. T1 and T2 values for fibroids and the myometrium were determined by drawing an elliptical region of interest on the tissue using postprocessing interactive tools (Osirix DICOM viewer, Bernex, Switzerland). The same ROI was utilized for T1, T2, and perfusion measurements. Quantitative values were determined from the mean ROI values. Fibroid viability was determined from the radiologist reading of in vivo contrast‐enhanced MR images and/or the official surgical pathology report. This processing included (H&E) staining of tissue samples that were used to assess tissue viability.

2.I. Acoustic property determination

Previous studies²⁸ have shown that tissue degradation was observed histologically 8 h after tissue excision. This study was designed such that all excised tissue was returned to the pathology laboratory within that time frame to minimize potential tissue degradation. In addition, pathology lab staffing required excised tissue to be returned by 20:00 on the study date. Therefore, it was possible to obtain tissue samples for acoustic property characterization when reperfusion, ex vivo MRI and thermal tissue characterization was completed within those time constraints. If the hysterectomy procedure occurred later in the day, these constraints limited the availability of tissue samples for acoustic properties measurement. This was the case for three uterus tissue samples and three fibroids. For the remaining cases, fresh tissue samples of the dissected specimens were obtained from the gross pathology lab to be used for acoustic property characterization on the following day. This included displaced‐water measurements for density, through‐transmission measurements for determining the speed of sound of the myometrium and fibroids,^{32, 33} and radiation force balance‐generated insertion‐loss measurements of acoustic attenuation.^{34, 35, 36} Based on previous experimental studies using these techniques, expected accuracy of the acoustic property measurements was 0.1 and 16% for speed of sound and acoustic attenuation measurements, respectively.³⁷

For all the measured properties in this study, differences between fibroid and myometrium and in vivo and ex vivo measurements were evaluated with a two‐tailed Student t test, using a p‐value of 0.05 to indicate statistically significant differences.

3. Results

Table 2 summarizes all the property measurements from the study with values presented as the mean ± 1SD. For fibroids, the N value in Table 2 reports the total number of fibroids evaluated with each measurement. For myometrium, the N value reports the total number of patients evaluated for each measurement. For all tissues, the n value indicates the total number of measurements obtained from the tissue samples.

Table 2.

Summary of all property measurements. For the myometrium and fibroids, N represents the number of patients and number of unique fibroids (several patients had more than one fibroid). The number of total measurements for all samples (n) is given for each property. The mean value and standard deviation from multiple samples is reported

General	Property	Units	Fibroids (N/n)		Myometrium (N/n)
	Fibroid size	cm	3.2 ± 2.3	(22/22)	‐‐	‐‐
MR properties	T2: in vivo	ms	43 ± 6	(10/10)	56 ± 9	(6/6)
	T2: ex vivo	ms	61 ± 15	(17/17)	69 ± 21	(8/8)
	T1: in vivo	ms	1201 ± 236	(10/10)	1426 ± 165	(6/6)
	T1: ex vivo	ms	1158 ± 193	(18/18)	1341 ± 284	(9/9)
Thermal properties	Thermal Diffusivity	mm2/s	0.217 ± 0.047	(11/49)	0.204 ± 0.039	(5/20)
	Thermal Conductivity	W/m/°C	0.538 ± 0.041	(11/49)	0.543 ± 0.050	(5/20)
	Density	kg/m3	1060 ± 16	(3/6)	1069	(1/1)
Acoustic properties	Speed of sound	m/s	1587 ± 18	(6/36)	1589 ± 23	(3/28)
	Attenuation coefficient	Np/cm/MHz	0.092 ± 0.021	(4/23)	0.052 ± 0.023	(1/7)

The individual fibroid and myometrium properties that were compiled in Table 2 are presented in Tables 3 and 4, respectively. When the same measurement was repeated multiple times in a tissue sample, results are presented as the mean ± 1SD with the number of measurements in parentheses. Specific notes explaining property measurements that were not acquired are included in Tables 3 and 4.

Table 3.

Individual fibroid property measurements. Depending on the fibroid size, excised tissue samples were either provided as a whole sample (WS) or morcellated (M). Fibroid size was determined from imaging when possible. When imaging was not available, the size reported was from the surgical pathology report (*)

Pt.	Size [cm]	Viability	Excision status	MR properties			Thermal properties			Acoustic properties		Notes
				T2wSI	T2 in vivo/ex vivo (ms)	T1 in vivo/ex vivo (ms)	Diffusivity (mm2/s)	Conductivity (W/m/°C)	Density (kg/m3)	Speed of sound (m/s)	Attenuation coefficient (Np/cm/MHz)
1	3.8*	Viable	WS	Hypo	‐‐/‐‐	‐‐/1179	0.165 ± 0.006 (n = 4)	0.546 ± 0.030 (n = 4)	‐‐	‐‐	‐‐	Patient not eligible for in vivo MRI
2	1.3 × 1.2 × 1.3	Viable	WS	Hypo	47/70	1070/1370	‐‐	‐‐	‐‐	‐‐	‐‐	Tissue sample too small for property measurements.
2	2.3 × 2.7 × 3.0	Viable	WS	Hypo	40/76	1010/1230	0.173 ± 0.019 (n = 3)	0.431 ± 0.067 (n = 3)	‐‐	‐‐	‐‐	Tissue not available for acoustic property measurements.
4	1.5 × 1.0 × 0.8	Viable	WS	Iso	‐‐/81	‐‐/1192	‐‐	‐‐	‐‐	1552 ± 2 (n = 6)	‐‐	Patient declined in vivo MRI. Tissue sample too small for property measurements.
5	4.9 × 5.1 × 4.2	Viable	M	‐‐	‐‐/‐‐	‐‐/‐‐	0.189 ± 0.024 (n = 5)	0.569 ± 0.010 (n = 5)	1065	1585 ± 4 (n = 5)	‐‐	Unable to schedule in vivo MRI before scheduled surgery. Tissue not available for acoustic measurements.
5	11.5 × 8.7 × 10.0	Viable	M	‐‐	‐‐/‐‐	‐‐/‐‐	0.185 ± 0.034 (n = 7)	0.505 ± 0.060 (n = 7)	‐‐	1590 ± 2 (n = 7)	‐‐
7	4.6*	Viable	WS	Hypo	45/68	1344/1060	0.293 ± 0.037 (n = 5)	0.566 ± 0.006 (n = 5)	1042	‐‐	‐‐	Tissue sample too small for property measurements.
8	2.1*	Viable	WS	Iso	49/52	1310/1231	‐‐	‐‐	‐‐	‐‐	‐‐	Fibroid not clearly accessible for thermal property measurements. Tissue sample too small for property measurements.
9	2.0*	Necrotic	M	‐‐	‐‐/53	‐‐/1250	0.281 ± 0.125 (n = 3)	0.581 ± 0.031 (n = 3)	‐‐	‐‐	‐‐	Patient declined in vivo MRI. Tissue sample too small for property measurements.
9	2.0*	Viable	M	‐‐	‐‐/48	‐‐/1299	0.224 ± 0.011 (n = 2)	0.557 ± 0.033 (n = 2)	‐‐	‐‐	‐‐
9	3.8*	Viable	M	‐‐	‐‐/47	‐‐/1065	0.170 ± 0.017 (n = 3)	0.559 ± 0.028 (n = 3)	‐‐	‐‐	0.097 ± 0.001 (n = 2)	Patient declined in vivo MRI.
10	3.1 × 4.0 × 5.5	Viable	M	Hyper	‐‐/91	‐‐/1451	0.221 ± 0.029 (n = 3)	0.544 ± 0.021 (n = 3)	‐‐	‐‐	‐‐	Subserosal hyperintense fibroid not identified during in vivo MRI for T1/T2 measurements. Tissue sample too small for property measurements.
10	4.0 × 3.8 × 4.0	Necrotic	M	Hypo	30/28	675/543	0.278 ± 0.082 (n = 4)	0.529 ± 0.024 (n = 4)	‐‐	‐‐	‐‐	Tissue sample too small for property measurements.
10	2.2 × 2.4 × 2.8	Viable	M	Hypo	43/59	1393/1216	‐‐	‐‐	‐‐	‐‐	‐‐
10	2.9x × 2.7 × 3.1	Viable	M	Hypo	46/58	1243/1184	‐‐	‐‐	‐‐	‐‐	‐‐
11	2.0 × 1.5 × 1.2	Viable	WS	Hypo	40/48	1168/1144	‐‐	‐‐	‐‐	‐‐	‐‐
11	1.0 × 1.0 × 1.2	Viable	WS	Hypo	‐‐/66	‐‐/972	‐‐	‐‐	‐‐	‐‐	‐‐
11	1.0 × 1.8 × 1.5	Viable	WS	Hypo	‐‐/73	‐‐/1023	‐‐	‐‐	‐‐	‐‐	‐‐
12	5.0*	Viable	M	‐‐	‐‐	‐‐	0.210 ± 0.041 (n = 10)	0.535 ± 0.030 (n = 10)	1074 ± 9 (n = 4)	1597 ± 12 (n = 6)	0.110 ± 0.013 (n = 12)	Patient declined in vivo MRI. Ex vivo MRI not obtained due to scheduling difficulties.
13	‐‐	Viable	M	‐‐	‐‐/71	‐‐/963	‐‐	‐‐	‐‐	1598 ± 2 (n = 7)	0.061 ± 0.004 (n = 5)	Unable to schedule in vivo MRI before scheduled surgery. Tissue not available for thermal property measurements.
13	‐‐	Viable	M	‐‐	‐‐	‐‐	‐‐	‐‐	‐‐	1601 ± 1 (n = 5)	0.100 ± 0.010 (n = 4)
14	2.3 × 2.5 × 2.3	Viable	WS	Iso	51/59	1505/1216	‐‐	‐‐	‐‐	‐‐	‐‐	Tissue sample too small for property measurements.
14	1.5 × 1.5 × 2.1	Viable	WS	Hypo	37/‐‐	1289/‐‐	‐‐	‐‐	‐‐	‐‐	‐‐	No ex vivo T1/T2 due to MRI acquisition error. Tissue sample too small for property measurements.
14	1.8 × 1.9 × 1.7	Viable	WS	Iso	‐‐/55	‐‐/1219	‐‐	‐‐	‐‐	‐‐	‐‐	Tissue sample too small for property measurements.

Table 4.

Uterus tissue property measurements. All samples were whole. No morcellated uterus tissue samples were processed (Participants 5, 9, 12, and 13)

Participant	MR properties		Thermal properties			Acoustic properties		Notes
	T2in vivo/ex vivo [ms]	T1 in vivo/ex vivo [ms]	Diffusivity [mm2/s]	Conductivity [W/m/°C]	Density [kg/m3]	Speed of sound [m/s]	Attenuation coefficient [Np/cm/MHz]
1	‐‐/‐‐	‐‐/1909	‐‐	‐‐	‐‐	‐‐	‐‐	Not eligible for in vivo MRI. Tissue not available for property measurement.
2	57/110	1270/1720	0.177 ± 0.008 (n = 3)	0.455 ± 0.017 (n = 3)	‐‐	‐‐	‐‐	Tissue not available for property measurement.
3	‐‐/57	‐‐/1295	‐‐	‐‐	‐‐	1616 ± 10 (n = 9)	‐‐	Unable to schedule in vivo MRI. Tissue sample too small for property measurement.
4	‐‐/71	‐‐/1077	0.175 ± 0.012 (n = 5)	0.578 ± 0.008 (n = 5)	‐‐	1575 ± 5 (n = 12)	‐‐	Tissue not available for property measurement.
6	‐‐/‐‐	‐‐/‐‐	0.189 ± 0.014 (n = 5)	0.553 ± 0.016 (n = 5)	1069	1577 ± 1 (n = 7)	0.052 ± 0.023 (n = 7)	Unable to schedule in vivo MRI.
7	58/81	1399/1145	‐‐	‐‐	‐‐	‐‐	‐‐	Tissue sample too small for property measurement.
8	42/47	1218/1228	0.207 ± 0.049 (n = 5)	0.572 ± 0.019 (n = 5)	‐‐	‐‐	‐‐
10	51/54	1517/1156	‐‐	‐‐	‐‐	‐‐	‐‐
11	69/79	1660/1351	0.270 ± 0.023 (n = 2)	0.557 ± 0.020 (n = 2)	‐‐	‐‐	‐‐
14	56/56	1493/1187	‐‐	‐‐	‐‐	‐‐	‐‐

The challenges in acquiring all tissue property measurements are highlighted in Tables 3 and 4. Six patients did not receive an in vivo MRI before their hysterectomy. This was either due to ineligibility, MRI scheduling conflicts, or the patient declining a presurgical MRI. For postoperative acoustic property analysis, in several cases either the fibroid was too small, or the tissue processing done during gross pathological analysis resulted in tissue samples too small. This occurred frequently, in 10 fibroid and 6 uterus tissue samples. In addition, while the uterus tissue was prepared to contain myometrium only, there is a possibility that endometrium tissue was present in some samples. When tissue samples were morcellated, it was impossible to definitively identify a sufficiently large tissue sample to measure uterus properties.

Of the 16 fibroids available for preoperative MRI testing, 69% demonstrated classic hypo‐intensity relative to the myometrium, with the remainder presented with iso‐ (25%) or hyper‐intensity (6%), as assessed by an expert observer (M.R.). This was further confirmed by comparing fibroids with the myometrium. Fibroid in vivo T2 values were significantly lower than the myometrium values (−23.2%, P = 0.005), indicating that most fibroids evaluated in the study demonstrated classic hypo‐intensity relative to the myometrium. T1 values measured in the ex vivo tissue were 3.6% and 6.0% lower than the in vivo values for the fibroid and myometrium tissues, respectively, but this difference was not statistically significant (P > 0.05). T2 measured in the ex vivo fibroid tissue was significantly higher than the in vivo measurement with an increase of 41.9% (P = 0.002). The ex vivo T2 measurement in the myometrium was also increased (23.2%) when compared to the in vivo measurement, but this was not statistically significant.

Thermal diffusivity was not significantly different between fibroid and myometrium tissues (0.217 ± 0.047 and 0.204 ± 0.039 mm²/s, respectively). Thermal conductivity, density and speed of sound measurements were also not significantly different between fibroid and myometrium tissue. However, the acoustic attenuation in fibroid tissue was significantly higher (P < 0.0001) than myometrium (0.092 ± 0.021 and 0.052 ± 0.02 Np/cm/MHz, respectively).

Tissue viability was maintained in the ex vivo tissues through regulation of temperature, pressure, and pH, along with effective reperfusion of the uterus. During experimental monitoring of the ex vivo tissue, measured temperatures ranged from 32.5°C to 40.5°C, pressure values were 70–200 mmHg, and pH varied between 6.7 and 7.8. Some set point adjustments were required to maintain physiological levels of temperature and pH at 37°C and 7.4, respectively.

Figure 2 shows photographs of a uterus specimen taken before cannulation and reperfusion (left) and immediately following the experiment (right). The lightening of the tissue color evident in the postexperiment photograph demonstrates that blood has been cleared from the uterine tissue. Most importantly, both gross pathology and histology assessments following experiments consistently revealed no evidence of tissue degradation.

Figure 2.

Photographs taken (a) before and (b) after experimental reperfusion of the uterus specimen (Participant 3). The uterine arteries are cannulated with different gauge catheters (pink 20‐g and yellow 24‐g) based on arterial size and access [Color figure can be viewed at wileyonlinelibrary.com]

Hematoxylin and Eosin stained sections from 14 patients were reviewed. Two tissue samples showed evidence of minimal fixation artifact involving the endometrium. This artifact was manifest by focal retraction of endometrial glands away from the surrounding stroma [Fig. 3(a)]. All other compartments of the uterine corpus (myometrium, serosa), including areas with pathologic findings (leiomyomata, adenomyosis) appeared well fixed in all [Figs. 3(a) and 3(b)]. No other remarkable morphologic alterations were identified, demonstrating that the extracorporeal perfusion system used for this study can maintain tissue viability while investigating tissue properties without compromising required histopathological assessments following surgery.

Figure 3.

Hematoxylin and Eosin stains from uterine tissue after surgical excision and reperfusion, indicating successful maintenance of tissue viability. (a) The endometrial stroma is focally retracted away (areas of white clearing) from endometrial glands, in an example of fixation artifact (200X). (b) Well‐fixed leiomyoma (left side) and adjacent normal myometrium (right side) (100X). (c) Area of well‐fixed adenomyosis and surrounding myometrium (100X) [Color figure can be viewed at wileyonlinelibrary.com]

Figure 4 shows representative images from an in vivo MR exam from participant 8. The axial oblique images are aligned with the long axis of the uterus and show a 4.6‐cm fibroid in the right uterine fundus (marked with an asterisk). The fibroid appears hypointense on T2‐weighted images [Fig. 4(a)] with a quantitative T2 map [Fig. 4(b)] yielding values of 45 and 58 ms for the fibroid and myometrium, respectively. T1 values [Fig. 4(c)] were 1344 ms in the fibroid and 1399 ms in the myometrium. Contrast‐enhanced MR images [Fig. 4(d)] showed uniform enhancement of the uterus (143% increase in signal intensity) and fibroid (124% increase), indicating a viable, perfused fibroid.

Figure 4.

Representative in vivo (a) T2‐weighted, (b) T2 Map, (c) T1 Map, and (d) postcontrast T1‐weighted MR images along the long axis of the uterus (Participant 8). A 4.6‐cm fibroid is seen in the right fundus (* marker). MR orientation of the images is indicated on (a). The scale in (a) can be applied to (b)‐(d) [Color figure can be viewed at wileyonlinelibrary.com]

MR images corresponding to the uterus specimen in Fig. 4, but during ex vivo experiments, are presented in Fig. 5. Slices were positioned along the long axis of the uterus to mimic in vivo imaging. During surgery, the cervix was split, explaining why the tissue splays in the ex vivo images. The primary fibroid is again marked with an asterisk. T2‐weighted images [Fig. 5(a)] show a still hypointense primary fibroid (* marker). T2 [Fig. 5(b)] in the fibroid and myometrium increased from the in vivo exam to 68 and 81 ms, respectively, and T1 values [Fig. 5(c)] decreased to 1060 and 1145 ms, respectively. Uniform enhancement of the uterus (9% increase in signal intensity) and fibroid (21% increase) in postcontrast MR images [Fig. 5(d)] demonstrates effective reperfusion. The background surrounding the tissue is nonuniform as water‐filled bags were used to position the specimen in the perfusate bath.

Figure 5.

Ex vivo (a) T2‐weighted, (b) T2 Map, (c) T1 Map, and (d) postcontrast T1‐weighted MR images along the long axis of the uterus. This is the same uterus specimen (Participant 8) shown in Fig. 3 with approximately the same orientation, but the cervix was split during surgery, explaining the splayed tissue at the bottom of these images. The 4.6‐cm fibroid is again marked with an asterisk in each image. The uterus is positioned with water‐filled bags surrounded by perfusate fluid. The scale in (a) can be applied to (b)–(d) [Color figure can be viewed at wileyonlinelibrary.com]

4. Discussion

Knowledge of patient‐specific treatment properties can inform and potentially improve all aspects of MRgFUS treatments, particularly treatment planning and MR sequence optimization for treatment monitoring and assessment. The variable response of fibroids to MRgFUS treatment is well documented in the literature. The required sonication power for ablation can vary dramatically both between patients and at different locations within a fibroid.^{38, 39, 40} Several retrospective studies have shown that fibroids with brighter T2‐weighted signal intensity (T2wSI) are more difficult to ablate,^{9, 16, 40, 41, 42, 43} resulting in reduced nonperfused volume (NPV) ratios immediately following treatment^{12, 40, 41, 42, 43, 44} and poorer long‐term outcomes,^{40, 41, 42, 43, 44, 45} High T2wSI of fibroids can be attributed to increased vascularization, fluid‐rich tissues, high cellularity, and/or degeneration,^{3, 46, 47, 48, 49, 50} so development of a technique that can accurately measure tissue characteristics and properties is of substantial use to the research community. This paper sought to meet this need by directly measuring patient‐specific MR, thermal and acoustic property values for both fibroid and myometrium tissues in both in vivo and ex vivo environments. While the data sets obtained from each tissue sample ranged in completeness, this set of measured properties represents a step forward in availability of patient‐specific properties. However, even though the ex vivo testing was designed to mimic the in vivo environment, ex vivo property measurements may vary from the in situ condition.

As reported by other works that have performed ex vivo uterine perfusion,^{28, 51} some uterus specimens are more difficult to perfuse due to variations in uterine vasculature and amount of vasculature cauterization after surgical excision. The MR‐compatible extracorpeal perfusion technique used for this study did not use any feedback control systems to monitor the temperature, pressure and pH values in the perfusate and reperfused tissue. All parameters were monitored manually and adjusted as needed to maintain appropriate value ranges and tissue viability. Temperature and pH variability during reperfusion should have minimal effect on thermal diffusivity, as this is largely driven by the water content in the tissue. Pressure variability could have a larger impact on water content and by extension, the thermal diffusivity, density, MRI and acoustic properties. However, in this study, once the tissue was reperfused, an increase of flow rate and pressure did not make the tissue sample visually swell in volume. Unfortunately, the sample size present in the study does not allow a more in depth study of how temperature, pressure and pH variability affect the presented tissue property measurements. The use of feedback control would possibly tighten the range of allowed values, resulting in less variability in tissue property measurements and extending tissue longevity.

The level of enhancement seen in the uterus and fibroid tissues during contrast‐enhanced MRI scans was decreased in the ex vivo environment when compared to the in vivo measurements. A 143% increase in signal intensity was seen in the uterus in vivo with only a 9% increase ex vivo, while a 124% increase was seen in the fibroid in vivo compared to 21% ex vivo. This discrepancy in signal enhancement is likely due to the ex vivo perfusion setup. While the setup was a closed loop system (Fig. 1), the amount of contrast uptake was likely limited due to the single bolus of contrast administered, the overall flow rate, varying pressure levels and circulation time.

One key limitation of the study is the small number of acoustic property measurements. As seen in Table 2, the number of samples used for acoustic property measurements is less than those used for thermal property measurements for both fibroid and myometrium tissues. This was due to both the study constraints and fibroid sizes. In general, smaller fibroids did not have sufficient volume to meet the needs of both clinical processing and our study. Study constraints caused acoustic property measurements to be obtained the day following the hysterectomy procedure. Even though the tissue samples were refrigerated overnight in degassed saline, tissue degradation likely occurred with an increase of gas present in the tissue. This could potentially cause an overestimation of the acoustic attenuation.

The number of thermal property measurements obtained from each sample was variable, constrained by both the study schedule and the size of the tissue available for testing. Larger samples collected earlier in the day facilitated more unique measurement locations. Generally, morcellated tissues had more fibroid tissue to assess. The smaller, intact fibroids allowed thermal property measurement if there was 1 cm of tissue on all sides of the probe, as recommended by the manufacturer. While almost all measurements between fibroids were consistent, one fibroid from patient 2 had a significantly lower mean thermal conductivity of 0.431 W/m/K. The source of this variability is unknown.

Despite the limitations in sample sizes and study constraints, the tissue properties measured during this study agree with other values in the literature. Olsrud et al.⁵² measured the thermal conductivity of excised myometrial tissue to be 0.536 W/m/K directly following hysterectomy, compared to 0.543 W/m/K measured in this study. In another study,²³ acoustic attenuation of excised myometrium tissue stored in saline 2 h prior to the measurement was determined using a through‐transmission technique. Values measured ranged from 0.058 to 0.18 Np/cm/MHz, similar to the 0.052 Np/cm/MHz mean uterus tissue value found in this work. A wide variability exists with both ex vivo and in vivo measured values, indicating the wide range of tissue property values that can exist in fibroids and highlighting the need for patient‐specific techniques to assess these properties. While all these studies have measured a single property value for the tissue sample, any fibroid could potentially have substantial heterogeneity within the sample. Therefore, assigning a single property value to a fibroid would likely result in varying levels of treatment prediction accuracy. This potential additional variability should be considered when using tabular tissue property values.

This extracorpeal perfusion technique could be applied for other tissue types if fresh tissue could be obtained. While similar tissue property measurement techniques have been performed on animal tissues,^{53, 54} a noninvasive in vivo technique would be preferred. Work is ongoing to develop methods to noninvasively measure both thermal and acoustic properties.^{25, 31, 55}

5. Conclusions

The presented protocol for the characterization of patient‐specific uterine fibroid MR, thermal, and acoustic properties has potential for improving MRgFUS ablation of uterine fibroids. The presented data increases the available database of properties for modeling computations used in MRgFUS treatment planning, monitoring, and control. The unique tissue characterization technique could be extended to other fresh tissue preparations expanding the indications that would benefit from this type of analysis.

Conflicts of interest

The authors have no relevant conflicts of interest to disclose.