Noriaki Kurimoto, MD; Masaki Murayama, MD; Shinkichiro Yoshioka, MD; Takashi Nishisaka, MD; Kouki Inai, MD; and Kiyohiko Dohi, MD

Transcription

1 Assessment of Usefulness of Endobronchial Ultrasonography in Determination of Depth of Tracheobronchial Tumor Invasion* Noriaki Kurimoto, MD; Masaki Murayama, MD; Shinkichiro Yoshioka, MD; Takashi Nishisaka, MD; Kouki Inai, MD; and Kiyohiko Dohi, MD Study objective: We assessed the usefulness of endobronchial ultrasonography in the determination of the depth of tumor invasion of the tracheobronchial wall. Methods: We performed a needle-puncture experiment on normal tissue of 45 specimens to determine the laminar structure of the tracheobronchial wall. In addition, we compared the ultrasonographic determinations of tumor invasion from 24 lung cancer cases with the histopathologic findings. Results: The cartilaginous portions of the extrapulmonary bronchi and the intrapulmonary bronchi exhibited a five-layer structure. Starting on the luminal side, the first layer (hyperechoic) was a marginal echo, the second layer (hypoechoic) was the submucosal tissue, the third layer (hyperechoic) was the marginal echo on the inner side of the bronchial cartilage, the fourth layer (hypoechoic) was bronchial cartilage, and the fifth layer (hyperechoic) was the marginal echo on the outer side of the cartilage. In the membranous portions, the first layer (hyperechoic) was a marginal echo, the second layer (hypoechoic) was smooth muscle, and the third layer (hyperechoic) corresponded to the adventitia. Comparisons between the ultrasonograms and the histopathologic findings in 24 lung cancer cases revealed that depth diagnosis was the same in 23 lesions (95.8%) and was different in 1 lesion (4.2%). In the single case in which the findings were different, lymphocytic infiltration that protruded between the cartilage rings was mistakenly interpreted as tumor infiltration. Conclusions: This method allows visualization of the laminar structure of the tracheobronchial wall, which is impossible with other diagnostic imaging methods. (CHEST 1999; 115: ) Key words: depth of tumor invasion; endobronchial ultrasonography; needle-puncture experiment; thin ultrasonic probe Abbreviation: EBUS endobronchial ultrasonography Many reports have shown that high-frequency ultrasound endoscopy is useful in determination of the depth of invasion of digestive system cancer. 1 5 Determination of the depth of invasion of tracheobronchial tumor lesions, predominantly squamous cell carcinoma, is the most important finding for determining the appropriate mode of therapy, whether it be local *From the Department of Surgery, Iwakuni Minami Hospital (Drs. Kurimoto and Murayama), the Second Department of Surgery (Drs. Yoshioka and Dohi) and the Department of Pathology (Dr. Inai), Hiroshima University School of Medicine, and the Department of Pathology, Hiroshima Prefectural Hospital (Dr. Nishisaka), Hiroshima, Japan. Manuscript received September 1, 1998; revision accepted January 12, Correspondence to: Noriaki Kurimoto, MD, Iwakuni Minami Hospital, Minami-Iwakunicho, Iwakuni City, Yamaguchi Prefecture, Japan laser destruction via bronchoscope or complete surgical resection. But CT and bronchoscopy, which have been used until now, have not been adequate, and predictions based on statistical probabilities using bronchoscopic measurements of the lesion have predominated. Since 1994, we have been performing endobronchial ultrasonography (EBUS) with a thinner ultrasonic probe inserted through the endoscopic working channel of a flexible bronchoscope and have been obtaining good images. In the present study, we conducted a needle-puncture experiment to identify the laminar structure by making comparisons between images of normal bronchial structure obtained by EBUS and the pathologic tissue. We then compared the EBUS images of resected lung cancer specimens and the histopathologic findings in completely sectioned speci Clinical Investigations

2 mens, and we assessed the usefulness of this method for determination of depth of tumor invasion. Materials and Methods Subjects The needle-puncture experiment was performed on the normal tissue of 45 specimens from human tracheas and bronchi that had been removed surgically for other clinical indications. The clinical cases of determination of depth of invasion involved 24 lung cancer cases in which the depth of invasion of the lesions in the tracheobronchial wall was determined using high-frequency ultrasonography between August 1994 and April The objective was to determine the accuracy of high-frequency ultrasound depth diagnosis compared with histopathologic findings after sectioning the entire specimen. Methods and Equipment Experiments were performed after informed consent was obtained. Needle-Puncture Experiment: Trachea and bronchi removed at surgery were cut into 1 1-cm pieces and were fastened to a rubber slab in two places at intervals of 1 cm with 23-gauge needles. Under a stereoscopic microscope a 29-gauge needle was inserted into the various layers from the cut end of the bronchus and was advanced so that it passed between the two 23-gauge needles in the long axis of the bronchial wall at right angles. The entire bronchial wall attached to the slab with the needles was submerged in water, and a probe was placed in the luminal side of the trachea or bronchus in the same direction that the 29-gauge needle was advanced. We scanned the specimen to obtain an image that included the two 23-gauge needles with a dot-like hyperechoic spot for the 29-gauge needle. For histopathologic evaluation, a cut was made perpendicular to the long axis of the bronchial wall that included the path of both 23-gauge needles, thus mimicking the view seen on the ultrasonogram. The hyperechoic spot of the 29-gauge needle on the ultrasonogram and the needle hole in the histopathologic specimens were compared to determine which layers in the pathologic tissue corresponded to the ultrasonographic laminar structures on the bronchial wall (Fig 1). Comparison of Clinical Surgical Specimens: Entire lobes of the lung, including the trachea and bronchi, that had been removed surgically were submerged in degassed water. The ultrasonic probe was inserted at the cut end of the trachea or bronchus. The probe was inserted as peripherally into the bronchus as possible, and, as it was slowly drawn toward the investigator (centrally), who was holding the probe in the center of the lumen, ultrasonograms were taken at right angles to the long axis of the trachea and bronchi, and the depth of tumor invasion was determined. After fixing the surgical specimen in formalin, sections were cut at 1-mm intervals perpendicular (round slices) to the long axis of the trachea and bronchi. The ultrasonographic determination of tumor invasion and the corresponding histopathologic findings were compared. Equipment In the study, we used a 20-MHz, radial mechanical-type ultrasonic probe (model UM-3R; Olympus; Tokyo, Japan) and an ultrasound unit (EU-M 20 Endoscopic Ultrasound System; Figure 1. Schema of needle-puncture experiment. A resected bronchial wall was fastened to a rubber slab with two 23-gauge needles. A 29-gauge needle was inserted into the various layers from the cut end and was advanced so that it passed between the two 23-gauge needles. We scanned the specimen to get an image that included the two 23-gauge needles with a dot-like hyperechoic spot of the 29-gauge needle. For the histopathologic evaluation of the scanning plane by ultrasound, a cut was made to include the path of both 23-gauge needles. The hyperechoic spot of the 29-gauge needle on the ultrasonogram and the needle hole in the histopathologic finding were compared to determine which layers in the pathologic tissue corresponded to the ultrasonographic layers. Olympus). The ultrasonograms were recorded with a printer (UP-880 Video Graphic Printer; Sony; Tokyo, Japan). Results Needle-Puncture Experiment The above-described needle-puncture experiment was performed on 45 specimens. A representative needle-puncture experiment specimen is shown in Figure 2. In a specimen in which the dot-like hyperechoic spot created by the 29-gauge needle was observed in the center of the outermost hypoechoic layer on the ultrasonogram of a segmental bronchus, the histopathologic findings showed a hole in the bronchial cartilage (Fig 2). This indicates that the outermost hypoechoic layer of the segmental bronchus was the cartilaginous layer. In another specimen in which the dot-like hyperechoic spot created by the 29-gauge needle was observed at the outer edge of the more luminal hypoechoic layer, the histopathologic findings showed the needle hole to be in the outermost side of the submucosal tissue (bronchial glands, smooth muscle), in contact with cartilage. This finding indicated that the hypoechoic layer on the luminal side corresponds to submucosal tissue (bronchial glands, smooth muscle), and the hypoechoic layer on the outside corresponds to bronchial cartilage. Furthermore, from the luminal side outward, the membranous portion consisted of hyperechoic, hypoechoic, and hyperechoic layers, CHEST / 115 / 6/ JUNE,

3 Figure 2. A representative specimen of the needle-puncture experiment. In a specimen in which the dot-like hyperechoic spot produced by the needles (black arrow) was observed in the center of the outermost hypoechoic layer (white arrow) of a segmental bronchus, the histopathologic finding showed a hole in the cartilage (black arrow), indicating that the outermost hypoechoic layer was the cartilage (bottom: hematoxylin-eosin, original 15). six cases in which the needle hole was in the bronchial cartilage, the dot-like hyperechoic spot of the needle was observed in the outermost hypoechoic layer (fourth layer). In all six cases in which the needle hole was in the adventitia, the dot-like hyperechoic spot of the needle was observed in the outermost hyperechoic layer (fifth layer). The membranous portion of the extrapulmonary bronchi is visualized as three layers. Fifteen specimens were compared using the membranous portion of the extrapulmonary bronchi. In the three cases in which the needle hole was in the submucosal tissue, the dot-like hyperechoic spot of the needle was observed in the most luminal hyperechoic layer (first layer). In the five cases in which the needle hole was in the smooth muscle, the dot-like hyperechoic spot of the needle was observed in the hypoechoic layer (second layer). In the seven cases in which the needle hole was in the adventitia, the dot-like hyperechoic spot of the needle was observed in the outermost hyperechoic layer (third layer). Conducting this experiment on the 45 specimens yielded the following results. The cartilaginous portion of the trachea and the extrapulmonary bronchi, as well as of the intrapulmonary bronchi, are visualized as five layers (Fig 3, Left). Starting on the luminal side, the first layer (hyperechoic) is a mar- and, in a specimen in which the dot-like hyperechoic spot created by the 29-gauge needle was observed in the second layer (the hypoechoic layer), the needle hole was found in the smooth muscle in the histopathologic specimen. This shows that the hypoechoic second layer corresponded to the smooth muscle layer. As a result of 45 needle-puncture experiments, the position of the needle holes in the pathologic tissue and the layers where the needles were located on the ultrasonographic images were compared. The intrapulmonary bronchi and the cartilaginous portion of the extrapulmonary bronchi are visualized as five layers. Thirty specimens were compared using the intrapulmonary bronchi and cartilaginous portion of the extrapulmonary bronchi. In 14 of the 18 cases in which the needle hole was in the submucosal tissue, the dot-like hyperechoic spots of the needle were observed in the more luminal hypoechoic layer (second layer), and in 4 cases, they were contained in the most luminal hyperechoic layer (first layer). In all Figure 3. Layers of the bronchial wall by EBUS. The cartilaginous portion of the trachea and the extrapulmonary bronchi is visualized as five layers, and the membranous portion is visualized as three layers (Left). The first layer (hyperechoic) is a marginal echo, the second layer (hypoechoic) is smooth muscle, the third layer (hyperechoic) is the marginal echo on the inner side of the bronchial cartilage, the fourth layer (hypoechoic) is cartilage, and the fifth layer (hyperechoic) is the marginal echo on the outer side of the cartilage. In the membranous portion of the extrapulmonary bronchi, the first layer (hyperechoic) is a marginal echo, the second layer (hypoechoic) is smooth muscle, and the third layer (hyperechoic) is the adventitia. The intrapulmonary bronchi are visualized as five layers (Right). The first layer (hyperechoic) is a marginal echo, the second layer (hypoechoic) is submucosal tissue, the third layer (hyperechoic) is the marginal echo on the inner side of the bronchial cartilage, the fourth layer (hypoechoic) is cartilage, and the fifth layer (hyperechoic) is the marginal echo on the outer side of the cartilage Clinical Investigations

4 ginal echo containing the epithelium and initial part of the submucosal tissue, the second layer (hypoechoic) is outermost submucosal tissue, the third layer (hyperechoic) is the marginal echo on the inner side of the bronchial cartilage, the fourth layer (hypoechoic) is bronchial cartilage, and the fifth layer (hyperechoic) is the marginal echo started at the outer side of the bronchial cartilage and contains the adventitia. In the membranous portion of the extrapulmonary bronchi, the first layer (hyperechoic) is a marginal echo containing the epithelium and the initial part of the submucosal tissue, the second layer (hypoechoic) is smooth muscle, and the third layer (hyperechoic) is the adventitia. In the intrapulmonary bronchi (Fig 3, Right), the first layer (hyperechoic) is a marginal echo, the second layer (hypoechoic) is submucosal tissue, the third layer (hyperechoic) is the marginal echo on the inner side of the bronchial cartilage, the fourth layer (hypoechoic) is cartilage, and the fifth layer (hyperechoic) is the marginal echo on the outer side of the cartilage. In the segmental bronchi and beyond, as the cartilage plates become progressively incomplete, the third and fifth marginal echoes become unclear and the determination of tumor invasion becomes difficult. Comparison of the Depth of Tumor Invasion as Determined by Ultrasonogram vs Histopathologic Findings The pathologic tissue of 24 lesions showed the depth of tumor invasion to be in the submucosal tissue in 7 lesions, bronchial cartilage in 1 lesion, adventitia in 1 lesion, and beyond the wall in 15 lesions. The depth of tumor invasion as determined by the ultrasonogram and the histopathologic findings was the same in 23 of 24 lesions (95.8%), but it was overestimated on the ultrasonogram in the other lesion. The sole exception was a case of squamous cell carcinoma that was observed to have invaded to the submucosal tissue histopathologically, whereas on the ultrasonogram there was a hypoechoic protrusion extending from between the cartilages to the adventitia. This was diagnosed as tumor invasion of the adventitia, but pathologically the entire hypoechoic region was a lymphocytic infiltration and, thus, the depth of tumor invasion had been overestimated. The comparative findings in 3 of the 24 lesions in which depth invasion was the same will be described as representative examples: 1. A squamous cell carcinoma was removed from the right intermediate trunk. The ultrasound showed that this lesion penetrated to but not into the third hyperechoic layer (the marginal echo on the inner side of the bronchial cartilage), thus limiting it to the submucosal tissue (Fig 4). Histopathologic findings confirmed this ultrasound determination of the depth of tumor invasion. 2. In the second case, in which the tumor involved the membranous portion of the bronchi, the ultrasound showed the tumor to be in contact with the second layer (hypoechoic), corresponding to the inner surface of the smooth muscle (Fig 5). Histopathologic findings confirmed this ultrasound determination of the depth of tumor invasion. 3. In the last case, involving a large squamous cell carcinoma, the ultrasound showed tumor surrounding a small hyperechoic island thought to be a marginal echo of cartilage, with destruction and replacement of surrounding tissue by tumor (Fig 6). Histopathologic findings confirmed this ultrasound determination of the depth of tumor invasion. Figure 4. Comparison of ultrasonogram (top) and histopathologic findings (bottom) of a representative example of invasion at submucosal tissue. A lesion was on the luminal side in contact with the hyperechoic third layer (inner marginal echo of cartilage; white arrow), and a tumor in contact with the inner surface (white arrow) of the cartilage was observed histopathologically. Histopathologic findings confirmed this ultrasound determination of the depth of tumor invasion (bottom: hematoxylin-eosin, original 5). CHEST / 115 / 6/ JUNE,

5 Figure 5. Comparison of ultrasonogram (top) and histopathologic findings (bottom) of a representative example of invasion above the smooth muscle of the membranous portion. A lesion was on the luminal side in contact with the hypoechoic second layer, which corresponds to smooth muscle (black arrow), and this was confirmed on the histopathologic specimen (bottom: hematoxylin-eosin, original 5). Figure 6. Comparison of ultrasonogram (top) and histopathologic findings (bottom) of a representative example of invasion beyond the bronchial wall. An island-like area (white arrow) was thought to be cartilage within the tumor on the basis of the ultrasonographic findings. Cartilage having exactly the same shape was observed at the same location (white arrow) in the histopathologic specimen (bottom: hematoxylin-eosin, original 5). Discussion Until now, bronchoscopic diagnosis of tracheobronchial diseases has advanced in conjunction with technologic advances. Improvements in electronic scopes, diagnosis by bronchial wall autofluorescence, transbronchial puncture cytodiagnosis, and endobronchial ultrasound diagnosis are current topics. We have been developing this EBUS since 1994, and, to date, have examined 600 cases and reported its usefulness. The indications for this method are the following: (1) determination of tumor invasion of tracheobronchial lesions; (2) positional relationships with the pulmonary artery and veins, and diagnosis of tumor invasion of pulmonary hilar tumors; (3) visualization of peritracheal and peribronchial lymph nodes, and diagnosis of metastasis; and (4) localization and qualitative diagnosis (diagnosis of whether benign or malignant) of peripheral lung lesions. The most important indication, something that can only be achieved with this method, is determination of the depth of tumor invasion of tracheobronchial lesions. There are reports by Hurter and Hanrath 6 and Becker 7 on the bronchial laminal structure using high-frequency ultrasonography. Hürter and Hanrath 6 claimed the bronchial laminal structure contains four layers. Becker 7 claimed that the tracheobronchial wall is composed of seven layers. But in these articles by Hurter and Hanrath 6 and Becker 7, accurate comparisons of ultrasonograms with histopathologic findings, such as those in our needle-puncture experiment, were not performed. Because a correct understanding of the laminar structure is necessary to perform an accurate determination of the depth of tumor invasion, we performed the needle-puncture experiment and com Clinical Investigations

6 pared ultrasonograms of surgical lung cancer specimens with the histopathologic findings in clinical cases. The needle-puncture experiment showed that with the current 20-MHz radial-type probe, the cartilaginous portion of the extrapulmonary bronchi and the intrapulmonary bronchi is depicted as a five-layer structure and that the membranous portion of the extrapulmonary bronchi appears as a three-layer structure. The fourth layer (hypoechoic), which represents the cartilage of the cartilaginous portion of the extrapulmonary bronchi and the intrapulmonary bronchi, and the second layer (hypoechoic), which represents the smooth muscle of the membranous portion, can often be clearly pointed out, and we consider following these layers to be the key to correct determination of depth of tumor invasion. When the depth of invasion is shallower than the cartilage, the intact normal cartilage layer can be followed. When invasion is deeper than the cartilage, however, the cartilage stands out in the tumor, its shape changes because of tumor invasion, and many times it even remains behind like an island. Thus, it is important to make repeated comparisons with the histopathologic findings. A precondition for obtaining the precise laminar structure by ultrasound is to position the probe at the center of the lumen so that the ultrasound enters the bronchial wall at right angles, and this is confirmed when the first layer is a hyperechoic, not hypoechoic, marginal echo. A point to be borne in mind when identifying the laminar structure of the wall by ultrasound is that marginal echoes 8,9 (hyperechoic bands produced by many reverberations) occur wherever there is a change in tissue. Marginal echoes, visualized as hyperechoic, are observed between the lumen and the mucosal epithelium, between the submucosal tissue and cartilage, and between cartilage and the adventitia. Aibe 9 has reported that marginal echoes include the transitional tissue, and that they are high linear echoes that extend distally in the direction of propagation of the ultrasound waves. Moreover, we measured the thickness of the mucosa in pathologic findings and the thickness of the first marginal echo of 10 specimens. The thickness of the mucosa was 0.05 mm on average. The thickness of the first marginal echo was 0.68 mm, and thus, more than 10 times the thickness of the mucosa. With this information, the fact that the hyperechoic needle mark was in the first hyperechoic layer in 4 of the 18 cases in which the needle hole was observed in the submucosal tissue in our needle-puncture experiment is explained by the fact that the marginal echo (first layer) extends from the inner margin of the mucosal epithelium to the inner part of the submucosal tissue. The marginal echo of the third layer seems to extend from the luminal margin of the bronchial cartilage to the middle of the cartilage, and the marginal echo of the fifth hyperechoic layer extends from the outer margin of the bronchial cartilage to the adventitia (Fig 7). Comparisons of the ultrasonographic and histopathologic findings in this study revealed that the depth of tumor invasion of the bronchial wall could be finely divided into four levels by this method: submucosa, cartilage, adventitia, and invasion beyond the adventitia. When a lesion is present from the first layer (hyperechoic, marginal echo) to the second layer (hypoechoic, submucosa), and when the third layer (hyperechoic, marginal echo on the inside of the cartilage) can be clearly followed, a depth diagnosis of submucosa can be made. When the lesion is present from the first layer (hyperechoic, marginal echo) to the fourth layer (hypoechoic, cartilage), and when the fifth layer (hyperechoic, marginal echo on the outside of the cartilage) can be clearly followed, a depth diagnosis of cartilage can be made. If the fifth layer is irregular, a depth diagnosis of adventitia can be made. Finally, when there are wedge-shaped interruptions in the third, fourth, and fifth layers, as well as a small island of the third, fourth, and fifth layers within the lesion, a depth diagnosis of invasion beyond the adventitia can be made. Because the mucosal epithelium is included in the first layer (hyperechoic, marginal echo), we do not think that carcinoma in situ can be visualized. The problems with determination of the depth of tumor invasion by this method at the present time Figure 7. Comparison of ultrasonographic and histopathologic layers of bronchi. Marginal echoes include the transitional tissue, and they are high linear echoes that extend distally in the direction of propagation of the ultrasound waves. The marginal echo (first layer) extends from the inner margin of the mucosal epithelium to the inner part of the submucosal tissue. The marginal echo of the third layer seems to extend from the luminal margin of the bronchial cartilage to the middle of the cartilage, and the marginal echo of the fifth hyperechoic layer extends from the outer margin of the bronchial cartilage to the adventitia. CHEST / 115 / 6/ JUNE,

7 include: (1) poor definition on the luminal side of the cartilage at 20 MHz; (2) attenuation by the balloon; (3) difficult visualization at bronchial spurs; and (4) the need for a thick, flexible bronchoscope because of the balloon sheath. The following measures are currently being considered to deal with these problems: (1) examination at a higher frequency, for example, 30 MHz; (2) improvement of balloon quality; (3) diagnosis by making longitudinal slices of the bronchi with a three-dimensional ultrasound system; (4) and making the probe thinner. Not going beyond the cartilage during bronchoscopic treatment of localized lesions has been reported to contribute considerably to its success and safety. 10,11 We think that having achieved the ability to determine the depth of tumor invasion using EBUS, which until now has been estimated on the basis of lesion size and height, represents a major advance in bronchoscopic technology. Future clinical applications of this method include in vivo evaluation of the contraction of bronchial smooth muscle, 12 diagnosis of the invasion of the pulmonary artery and veins by hilar tumors, 13 localization of peripheral lung lesions, determination of whether a lesion is malignant or benign, 14 and diagnosis of mediastinal lymph node metastasis. 15,16 We are currently adding cases to the series, and even further development is expected. Conclusions Ultrasonograms obtained at 20 MHz showed five layers in the cartilaginous portion of the extrapulmonary and intrapulmonary bronchi, and three layers in the membranous portion. The key to determination of the depth of tumor invasion is to follow the fourth layer (hypoechoic), which represents bronchial cartilage, in the cartilaginous portion of the extrapulmonary and intrapulmonary bronchi, and to follow the second layer (hypoechoic), which corresponds to smooth muscle, in the membranous portion. Comparison of the determination of the depth of tumor invasion on the basis of ultrasonography and histopathologic findings in the surgical specimens of 24 lesions showed that the findings were the same in 23 lesions (95.8%) and different in only 1 lesion (4.2%). In the single case in which the depth of tumor invasion was different, lymphocytic infiltration protruding between the cartilages was mistakenly interpreted by ultrasound as tumor invasion. This area of lymphocytic infiltration was clearly depicted by ultrasound, but, in this case, the ultrasound was unable to differentiate lymphocytic infiltration from tumor invasion. The problems with this method at the present time are poor definition on the luminal side of the cartilage at 20 MHz, attenuation by the balloon, difficult visualization at bronchial spurs, and the need for a thick, flexible bronchoscope because of the balloon sheath. References 1 Grimm H, Binmoeller KF, Hamper K, et al. Endosonography for preoperative locoregional staging of esophageal and gastric cancer. Endoscopy 1993; 25: Abe S, Lightdale CJ, Brennan MF. The Japanese experience with endoscopic ultrasonography in staging of gastric cancer. Gastrointest Endosc 1993; 39: Rösch T, Classen M. Endoscopic ultrasonography. In: Hunter J, ed. Gastrointestinal endoscopy. 2nd ed. London, UK: PB Cotton, 1993; Rösch T. Endoscopic ultrasonography. Endoscopy 1992; 24: Murata Y, Muroi M, Yoshida M, et al. Endoscopic ultrasonography in diagnosis of esophageal carcinoma. Surg Endosc 1987; 1: Hürter T, Hanrath P. Endobronchial sonography: feasibility and preliminary results. Thorax 1992; 47: Becker H. Endobronchialer Ultraschall-Eine neue. Perspektive in der Bronchologie. Ultraschall in Med 1996; 17: Kimmey MB, Martin RW, Haggitt RC, et al. Histologic correlates of gastrointestinal ultrasound images. Gastroenterology 1989; 96: Aibe T. A study on the structure of layers of the gastrointestinal wall visualized by means of the ultrasonic endoscope: the structure of layers of the esophageal wall and the colonic wall. Gastroenterol Endoscop 1984; 26: Konaka C, Okunaka T, Kato H. Combined use of photodynamic therapy. Ann Thorac Cardiovasc Surg 1995; 1: Okunaka T, Kato H, Konaka C, et al. Photodynamic therapy for multiple primary bronchogenic carcinoma. Cancer 1991; 68: Iizuka K, Dobashi K, Houjou S, et al. Evaluation of airway smooth muscle contractions in vitro by high-frequency ultrasonic imaging. Chest 1992; 102: Hurter T, Hanrath P. Endobronchiale Sonographie zur Diagnostik pulmonaler und mediastinaler Tumoren. Dtsch Med Wochenschr 1990; 115: Goldberg BB, Steiner RM, Liu JB, et al. US-assisted bronchoscopy with use of miniature transducer-containing catheters. Radiology 1994; 190: Schuder G, Isringhaus H, Kubale B, et al. Endoscopic ultrasonography of the mediastinum in the diagnosis of bronchial carcinoma. Thorac Cardiovasc Surg 1991; 39: Kondo D, Imaizumi M, Abe T, et al. Endoscopic ultrasound examination for mediastinal lymph node metastases of lung cancer. Chest 1990; 98: Clinical Investigations