This is a 66-year-old woman who was evaluated for a suspected gastrinoma 18 months before this presentation. An octreotide scan showed a questionable area of increased activity anterior to the right kidney. A computed tomography (CT) scan and an endoscopic ultrasound (EUS) were performed at the time and nothing was found. The patient’s gastrin levels continued to rise and a second octreotide scan was performed. The report stated that the “same region of mild increased activity is seen medial and anterior to the right kidney.”

A repeat CT scan (previously reported as negative) was read as showing a “questionable tiny focal area of soft tissue density at the medial aspect of the second portion of the duodenum which measures 1.4 cm and corresponds to the octreotide abnormality, suspicious for a small gastrinoma in the gastrinoma triangle.” The corresponding slice of the CT is shown here; the arrow inserted by the radiologist can be seen.

The repeat EUS shows a lesion, which is considerably smaller (9.5 mm) and in a different location from what was seen on the CT scan. In fact, the lesion identified on the EUS is exactly medial and anterior to the right kidney (as would be expected based on the octreotide scan), whereas the lesion as postulated by the CT is more remote.

In retrospect, a lesion corresponding to the EUS image could also be identified on the CT, anterior and medial to the right kidney and located in the bowel wall. It appears to be slightly enhancing, as is typical for islet cell tumors [4].

Confirmation bias, or the willingness to identify abnormalities that are expected to be present but are really not, at least not where suspected, is a trap everybody can fall into. There is a blurry line between looking so hard that one starts seeing things and not looking hard enough. The next section will examine this trade-off in more detail.

Image perception research deals with the processes of perception and recognition of information in medical images. It is estimated that there is about a 20–30 % miss rate (false negatives) in radiology, with a 2–15 % rate of false positives [5]. Whether these numbers are really so high could be debated; let us agree, however, that they are not insignificant. Using eye position recording techniques, it has been found that false negatives can be classified into three categories based on how long they are fixated or dwelled on (visual dwell patterns) [5, 6].

Decision error can be illustrated by the following case and conceptualized by the cognitive theory of field dependence. A field-dependent person has difficulty finding a geometric shape that is embedded or “hidden” in a background with similar (but not identical) lines and shapes. The conflicting patterns distract the person from identifying the given figure. A person who is field-independent can readily identify the geometric shape, regardless of the background in which it is set [7]. However, identifying hidden figures has a flip side: subjective contour formation. Subjective contour formation is a psychophysiological phenomenon in which complete “structures” are falsely constructed from minimal outlines [8].

The EUS image recognition process relies heavily on patterns that correlate with abnormalities. Lack of awareness of additional patterns or finding atypical changes may confuse the endosonographer, who could either ignore the finding or make incorrect conclusions. In these cases an open mind and awareness of unknown possibilities should be recognized. Video-recording of these findings may be useful for additional review with radiologists or other endosonographers.

There are three domains of image resolution: temporal, detail (axial and lateral), and contrast. Detail resolution is largely determined by image frequency and beam width. Contrast resolution, on the other hand, is generally determined by pre- and postprocessing of images [10]. Postprocessing curves can be selected on the ultrasound processor and sometimes need to be adjusted to enhance small differences in echo intensity. Some endosonographers never change their postprocessing assignments. For beginners, staying with the scheme can lead to quicker familiarization with the strange world of sonography. However, the next case illustrates how not using postprocessing curve adjustments can lead to problems.

A 50-year-old man with a history of alcohol problems but no drinking for 6 years was admitted to the hospital with acute pancreatitis. He later developed jaundice. Endoscopic retrograde cholangiopancreatography (ERCP) showed a smoothly compressed bile duct. A stent was placed. CT and magnetic resonance imaging (MRI) were negative for a mass lesion. The CA 19-9 was initially elevated but this was thought to be due to the biliary obstruction and acute pancreatitis. EUS was performed; no mass was seen. Features of mild chronic pancreatitis, especially in the body and tail, were present. Three days after the first EUS the patient’s repeat CA 19-9 levels came back: they had quintupled. He was brought back for a second look. Linear EUS showed the tumor and biopsies were positive for adenocarcinoma. The linear electronically switched instrument is widely believed to have an intrinsically better contrast resolution than the mechanical radial instrument, but different postprocessing settings may have revealed the tumor on the first exam.

This is a 73-year-old patient with generally declining health who was admitted to the hospital with a diagnosis of cholecystitis. He was considered to have poor operative risk and instead of a cholecystectomy a cholecystomy was performed. The CT scan of the abdomen showed small pancreatic cysts, and EUS evaluation was requested. The EUS showed small simple-appearing cysts, which were ignored in view of the finding of a 6-cm-diameter periesophageal lymph node mass. Biopsies revealed Hodgkin’s lymphoma.

Prompted by the EUS findings, a CT scan of the chest was ordered. This CT was read as showing a 2.5-cm lymph node in the azygoesophageal recess. Interestingly, the 6-cm lymph node mass was not picked up by the reading radiologist, presumably in part because streak artifacts on the CT scan further deteriorated image quality in the region of interest.

This case illustrates two things. First, EUS is an imaging modality that looks at the same anatomy based on a reflection-based algorithm, whereas CT evaluates tissue based on its transmission characteristics. Therefore, the same pathology can be obvious on EUS and close to invisible on CT or vice versa. Second, radiologists and endosonographers need to communicate better.

Hounsfield units are a quantitative measurement of radiodensity on CT scans. Low Hounsfield units in the pancreas say little more than that the lesion is hypodense; it does not reliably distinguish between a hypodense solid tumor mass or a cyst, as this case demonstrates. Both lesions in this patient’s pancreas were thought to be cysts on CT on the basis of overall impression and density measurements. EUS shows that the lesion in the head is indeed a cyst, whereas the pancreatic tail lesion is adenocarcinoma, as proven by a biopsy.

Reverberation artifact in ultrasound imaging is caused by the multiple reflections of ultrasound back and forth between two closely spaced interfaces. It is commonly seen in the human body when air is trapped between tissue planes, which conduct sound well (soft tissue-to-gas interfaces). It occurs even more frequently in endosonography when the balloon around the ultrasound transducer is only partially filled with water and contains air bubbles. On the basis of its appearance, the artifact is also referred to as “ring-down artifact” and “comet tail artifact.”.

This is a patient with a cholangiocarcinoma. The patient had an ERCP with placement of a stent across his common bile duct stricture 4 days before this EUS. During the EUS the gallbladder initially appeared normal; however, eventually enough air was blown from the duodenum via the stent into the gallbladder to cause the reverberation artifact shown. Chapter 2, Esophagus and Mediastinum, shows how this artifact can help detect air-filled bronchi in a consolidated lung.

This is a 62-year-old man with a carcinoid of the terminal ileum metastatic to the liver. He is now being considered for debulking surgery. The CT scan showed a possible area where it seemed as if the large centrally located metastasis was invading into the portal vein. EUS evaluation was requested and did not show any such invasion. This study also illustrates the usefulness of the vascular Doppler capability of solid-state EUS scanners.

Partial volume artifacts occur with CT and MRI when a structure is only partly contained within the imaging section, voxel, or pixel. The effect is most marked in regions where two tissues with different signal intensities adjoin, such as in this case (the tumor and portal vein). In one plane this may lead to blurring, but examination of the image in another plane will often show clearly defined borders [11]. EUS is an imaging technique that creates its own artifacts (albeit of a different nature); however, when the operator suspects an artifact, the imaging plane can be adjusted and the question often resolved.

This is a 60-year-old woman who had a plethora of complaints and was referred for EUS for the evaluation of a cyst in the pancreatic tail. Upon withdrawal of the echoendoscope through the cervical esophagus a large thyroid was incidentally noted. An evaluation of thyroid-stimulating hormone levels was ordered and came back as <0.01 uIU/ml, consistent with hyperthyroidism, in this case most likely Graves’ disease. There is at least one case report where EUS was used to biopsy thyroid nodules [12], but it is unlikely that there is a role for EUS in the biopsy of thyroid masses. Percutaneous biopsies are well established, do not require sedation, and allow a view of the large portions of the thyroid, which are invisible (blocked by the trachea) to the endosonographer. Although not often tasked to evaluate the thyroid, endosonographers should be able to recognize obvious pathology. On ultrasound, thyroid tissue has a uniform solid pattern consisting of dense agglomerates of fine echoes of equal size. Compared to adjacent muscle the normal thyroid is hyperechoic. A hypoechoic appearance of the thyroid is most often due to an inflammatory process. Anteroposterior diameters of 2.0 cm or more are suspicious for pathology, and diameters exceeding 2.5 cm are clearly abnormal [13].

Esophageal and mediastinal anatomy are best learned together. In fact, they are inseparable. This point is emphasized in Chap.​ 2, where the readers are directed to further resources. The following case illustrates that sometimes “ad hoc” landmarks need to be used to make sense of imaging findings, and this often involves the integration of several different imaging studies with EUS.

This is a 73-year-old man with a newly diagnosed left upper lobe lung nodule. Unfortunately, transthoracic biopsies revealed adenocarcinoma. The positron emission tomography–CT scan shows this nodule; it also shows another hot spot suspicious for a metastatic mediastinal lymph node on the contralateral side. This would make the patient unresectable.

The patient also has a history of a resected esophageal cancer some 12 years earlier. EUS-guided biopsies of the suspected mediastinal lymph node were requested.

A bony spur identified on EUS as causing an esophageal impression was helpful in establishing the nature of the hot spot seen on positron emission tomography.

Anatomic landmarks in the upper abdomen are reviewed in the respective organ-specific chapters. Here we review pelvic structures that are not normally the focus of the gastrointestinal endosonographer.