MOORE, Circuit Judge.
Apple appeals from the final decision of the International Trade Commission (ITC) that the asserted claims of U.S. Patent No. 7,663,607 (’607 patent) are invalid and that Motorola does not infringe the asserted claims of U.S. Patent No. 7,812,828 (’828 patent). Apple challenges the ITC’s claim construction and its determinations of obviousness, anticipation, and noninfringement. For the following reasons, we affirm-in-part, reverse-in-part, and vacate-in-part the ITC’s decision and remand for further proceedings.
Background
This patent case involves smartphone touchscreens. The '607 patent discloses a touch panel with a transparent capacitive sensing medium that can detect multiple touches at once. '607 patent, at [57]. To achieve the multitouch functionality, the touch panel employs a matrix of electrodes connected to circuits that measure the change in charge that occurs as a result of pressure applied to the screen. Id. col.5 1.27-col.6 1.7. The pressure-induced change occurs because the electrode rows are in a different layer than the electrode columns. Id. col.5 1.15-col.6 1.18. When a user touches the screen, the pressure applied at each intersection point causes charge to flow between the electrodes at that node. Id. Measuring circuits connected to the electrodes scan the matrix and measure the displaced charge at each node. Id. By detecting these changes, the touch panel can determine if and where a user has touched the screen. Id.
The '607 patent also discloses how to make the touchscreen transparent. It teaches constructing the electrodes with [1360]*1360indium tin oxide (ITO), a transparent material. '607 patent, col. 12 11.35-52. But simply forming the electrodes from ITO may not render the matrix invisible because the ITO electrodes tend to be less transparent than gaps in the electrode matrix. Id. col.14 1.60-col.l5 1.23. To remedy this problem, the patent teaches the use of “dummy” ITO pads to fill in gaps in the matrix. Id. col.15 11.8-24. By inserting these pads in the matrix gaps, the matrix has the optical properties of a uniform sheet of ITO and thus becomes invisible to the user. Id.
The '828 patent discloses a method to determine if the displaced charge at the nodes corresponds to a finger touching the screen. It teaches that the touch panel software “mathematically fit[s] an ellipse” around the nodes at which the measuring circuits have detected a touch. '828 patent, figs. 13-15, col.60 1.5-16. Performing the “fit” allows the device to determine if pressure applied to the screen constitutes a finger touch as well as track the movement of the finger across the touchscreen. Id. at [57].
Apple initiated proceedings in the ITC, alleging that Motorola’s smartphones and tablets infringed various claims of the '607 and '828 patents. Apple alleged that Motorola infringed claims 1-7 and 10 of the '607 patent and claims 1, 2, 10, 11, 24-26, and 29 of the '828 patent. Claim 1 of the '607 patent is representative of the asserted touch panel claims:
A touch panel comprising a transparent capacitive sensing medium, configured to detect multiple touches or near touches that occur at a same time and at distinct locations ... wherein the transparent capacitive sensing medium comprises:
a first layer having a plurality of transparent first conductive lines ...; and a second layer spatially separated from the first layer and having a plurality of transparent second conductive lines ... each of the second conductive lines being operatively coupled to capacitive monitoring circuitry;
wherein the capacitive monitoring circuitry is configured to detect changes in charge coupling between the first conductive lines and the second conductive lines.
'607 patent, claim 1 (emphases added). Claim 10, also disputed on appeal, recites a similar display arrangement and requires the touch panel to form a “pixilated image.” Claim 1 of the '828 patent is representative of the asserted claims relating to mathematically fitting an ellipse:
A method of processing input from a touch-sensitive surface, the method comprising:
receiving at least one proximity image representing a scan of a plurality of electrodes of the touch-sensitive surface; segmenting each proximity image into one or more pixel groups that indicate significant proximity, each pixel group representing proximity of a distinguishable hand part or other touch object on or near the touch-sensitive surface; and mathematically fitting an ellipse to at least one of the pixel groups.
'828 patent, claim 1 (emphasis added). Motorola prevailed in the ITC proceedings. While the ITC determined that an article describing SmartSkin, a prior art touchscreen system, did not anticipate the asserted claims of the '607 patent, it determined that SmartSkin rendered those claims obvious. The ITC also found that U.S. Patent No. 7,372,455 (Perski '455) anticipated the '607 patent claims. The ITC also found that Motorola did not infringe the '828 patent. It construed the term “mathematically fitting an ellipse” to require the method to perform “a mathematical process” whereby “an ellipse is actually fitted to the data.” J.A. 58-70. [1361]*1361Finding that the Motorola products do not fit an ellipse to the electrode data, the ITC determined that those products do not infringe the asserted claims of the '828 patent.
Apple appeals. We have jurisdiction under 28 U.S.C. § 1295(a)(6).
Discussion
I. Standard of Review
We review the ITC’s legal determinations de novo and its factual findings for substantial evidence. Crocs, Inc. v. Int’l Trade Comm’n, 598 F.3d 1294, 1302 (Fed.Cir.2010). Claim construction is a matter of law, which we review de novo. Sorensen v. Int’l Trade Comm’n, 427 F.3d 1375, 1378 (Fed.Cir.2005). Obviousness is a question of law based on underlying facts. Crocs, 598 F.3d at 1308. We review the ITC’s obviousness determination without deference and its factual findings for substantial evidence. Id. Whether a prior art reference anticipates the claims is a question of fact, reviewed for substantial evidence. Vizio, Inc. v. Int’l Trade Comm’n, 605 F.3d 1330, 1342 (Fed.Cir.2010).
II. Anticipation of the '607 Patent: Perski '455
The ALJ found that Perski '455 anticipates the asserted claims of the '607 patent. He found that Perski '455 was § 102(e) prior art despite Apple’s allegation of conception prior to the fifing date of the application that issued as Perski '455. The ALJ found that the provisional application to which Perski '455 claims priority, U.S. Provisional Patent Application No. 60/446,808 (Perski '808), provides written description support for the disclosure in Perski '455. After resolving the priority issue against Apple, the ALJ determined that Perski '455 anticipates the '607 patent claims. The ALJ found that Perski '455 discloses a touchscreen that can detect multiple touches at the same time. The ITC declined to review these findings.
Apple argues that the ITC anticipation findings were in error. It contends that Perski '455 is not prior art because (1) Perski '808 does not disclose any way to determine whether multiple fingers touch the screen; and (2) Perski '808 does not specifically incorporate by reference the “front end” and “digital unit” aspects of U.S. Provisional Patent Application 60/406,662 (Morag) that the ALJ used to find claim 10 anticipated.
Even if Perski '455 is prior art, Apple argues that the reference does not disclose “detectpng] multiple touches or near touches that occur at a same time and at distinct locations.” It contends that the algorithm disclosed in Perski '455 cannot detect multiple touches that occur at the same time because it requires too much processing—the algorithm requires at least n*m steps to accurately scan all the nodes in a sensor matrix containing m rows and n columns. Apple asserts that Motorola also failed to present any evidence that the matrix disclosed in Perski '455 can accurately detect multiple touches at the same time because a single large touch can cause an output signal to be detected on more than one conductor line.
The ITC and Motorola respond that Perski '455 is prior art to the '607 patent. They argue that Perski '808 discloses the same sensor matrix and multitouch detection algorithms as Perski '455. Regarding claim 10, Motorola argues that Perski '808 specifically incorporates the relevant portions of Morag.
The ITC and Motorola argue that Perski '455 discloses all of the limitations of the '607 patent claims. They argue that Per-ski '455 discloses a sensor that can detect multiple touches at the same time. They contend that the claims do not require a particular speed or accuracy in detecting [1362]*1362the multiple touches, and regardless, Per-ski '455 discloses both “simple” and “faster” detection algorithms. Lastly, Motorola asserts that Perski '808 discloses the exact scanning method that the '607 patent discloses to “detect multiple touches or near touches that occur at a same time and at distinct locations.”
As an initial matter, we agree with the ITC and Motorola that substantial evidence supports the ITC’s determination that the disclosure in Perski '808 provides adequate written support for Perski '455. Perski '808 provides the same multitouch scanning algorithms as Perski '455. Both disclose a sensor matrix that senses a touch by scanning the nodes of the matrix. Both disclose a “simple and direct approach” in which the circuitry scans each node of the matrix, which requires at least n*m steps for a sensor matrix that contains n columns and m rows. Each reference also discloses the same “faster approach.” Specifically, each discloses scanning the nodes affiliated with a group of lines on one axis, which requires between two steps and n*m steps depending on the number of lines in the group. This faster approach, however, is not as accurate when detecting multiple touches that occur simultaneously at specific locations. To remedy this problem, both references disclose the “optimal approach” of combining the two methods to achieve the right balance of speed and accuracy. Thus, substantial evidence supports the ITC’s finding that Perski '808 provides written support for Perski '455.1
We agree with Apple, however, that Perski '808 fails to incorporate by reference Morag.2 For a prior art reference to anticipate a claim, the reference must disclose each claim limitation in a single document. Advanced Display Sys., Inc. v. Kent State Univ., 212 F.3d 1272, 1282 (Fed.Cir.2000). The prior art document, however, may incorporate subject matter by reference to another document such that the incorporated material becomes part of the host document for the purposes of anticipation. Id. “To incorporate material by reference, the host document must identify with detailed particularity what specific material it incorporates and clearly indicate where that material is found in the various documents.” Id. at 1282-83. Whether and to what extent a host document incorporates material by reference is a question of law, subject to de novo review. Id. at 1283.
Here, Perski '808 only makes a passing reference to Morag as a “method similar” for detecting the presence of a stylus. J.A. 16149. It does not affirmatively incorporate any information. Perski '808 does not even refer to the particular functionality in Morag that detects the presence of a stylus, let alone the process that outputs touch event information to form a [1363]*1363pixilated image, as required by claim 10. Thus, Perski '808’s reference to Morag falls short of identifying with detailed particularity the material that discloses the “pixilated image” limitation in claim 10. Because Perski '808 does not incorporate by reference the anticipatory subject matter from Morag, the ITC’s finding that Perski '455 anticipates claim 10 of the '607 patent lacks substantial evidence.
Having resolved that Perski '455 is prior art for claims 1-7 of the '607 patent, we conclude that substantial evidence supports the ITC’s finding that Per-ski '455 anticipates those claims. Perski '455 discloses an embodiment in which “[a] two-dimensional sensor matrix lies in a transparent layer over an electronic display device” and a finger touch at a certain location on the matrix “increases the capacitance between the first conductor line and the orthogonal conductor line which happens to be at or closest to the touch position.” Perski '455, col.13 11.32-40 (reference numerals omitted). It discloses two matrix scanning algorithms that are “preferably able to detect more than one finger touch at the same time.” Id. col.14 11.15-19.
As recounted above, the number of steps required to scan the matrix depends on the number of matrix columns and rows. The matrix disclosed in Perski '455 has n columns and m rows. Perski '455, col.13 1.65-col.l4 1.4. Apple is correct that the slower method disclosed in Perski '455 requires at least n*m steps to scan the matrix because it scans each node one by one. Id. col.14 11.20-43. But the reference also discloses a “faster approach” that requires between two steps and a “maximum of n*m steps.” Id. col.14. 11.44-56. The faster approach scans groups of nodes per step, which significantly reduces the number of steps required to scan the matrix. Id. The reference also discloses a blend of the slower and faster approaches as the “optimal approach” to detecting multiple touches. Id. col.14 1.57-59.
Apple fails to provide any reason why the faster or optimal approaches would be too slow or inaccurate to detect multiple touches or why the disclosure of Perski '455 fails to enable multiple touches. To the contrary, as Motorola points out, the scanning algorithm disclosed in the '607 patent is very similar to the “faster approach” disclosed in Perski '455. The '607 patent discloses a sensing circuit that detects changes in capacitance at each node along n columns in the matrix by cycling through one row at a time for the m rows. '607 patent, col.5 1.60-eol.6 1.6. Moreover, the claims of the '607 patent do not expressly contain a speed or accuracy limitation. Thus, we conclude that substantial evidence supports the ITC’s finding that Perski '455 anticipates claims 1-7 of the '607 patent. The ITC’s decision that Per-ski '455 anticipates claim 10, however, lacks substantial evidence.
III. Anticipation and Obviousness of the '607 Patent: SmartSkin
A. Anticipation
Motorola argues that if we reverse the ITC’s decision that Perski '455 anticipates claim 10 of the '607 patent, we should reverse the ITC’s decision that SmartSkin does not anticipate claim 10. The ALJ determined that SmartSkin does not disclose the use of transparent conductive lines because the reference’s statements about using transparent ITO conductive lines related to future work. The ITC declined to review the ALJ’s finding. Motorola argues that the ALJ erred because SmartSkin’s disclosure would have enabled a skilled artisan to build a touchscreen using transparent ITO electrodes.
We agree with Apple and the ITC that substantial evidence supports the ITC’s finding of no anticipation. SmartS-[1364]*1364kin discloses an opaque surface covered with a grid of copper electrodes, not a transparent touchscreen based on ITO electrodes. In the SmartSkin system, a projector displays an image on the surface and circuitry connected to the copper electrode grid detects when a user touches the surface, enabling the surface to operate as a touch-screen. SmartSkin explains that its authors had developed two “working interactive surface systems based on this technology: a table and a tablet.” J.A. 13603. Figure 7 from SmartSkin shows an exemplary “table” system:
[[Image here]]
J.A. 13599. Thus, the reference explains that the authors had not achieved a touchscreen employing transparent electrodes.
The only discussion of transparent electrodes appears under the “Conclusions and Directions for Future Work” section, in which the authors explain that they were interested in future “research directions.” J.A. 13603. One of those directions was the use of transparent ITO electrodes that are “mounted in front of a flat panel display or a rear-projection screen.” Id. There is no disclosure that the authors had achieved a transparent touch screen and the record does not indicate that it would have been routine to do so. Nor is there any disclosure in SmartSkin that the matrix of ITO electrodes would have created the “transparent ... layerfs]” recited in claim 10. Although the ITO electrodes are transparent, the '607 patent explains that, when arranged in a matrix, “the patterned ITO can become quite visible thereby producing a touchscreen with undesirable optical properties.” '607 patent, eol.14 1.65— col.15 1.3.
We do not agree with Motorola that the ITC’s determination regarding the disclosure of the SmartSkin reference lacks substantial evidence. Given SmartSkin’s limited disclosure, we decline to disturb the ITC’s finding that Motorola failed to prove that SmartSkin anticipates claim 10 of the '607 patent.
B. Obviousness
Despite finding that SmartSkin did not anticipate the '607 patent claims, the ALJ concluded that they would have been obvious in light of SmartSkin in combination with a patent application that stemmed from the SmartSkin project, Unexamined Japanese Patent Application No.2002342033A (Rekimoto). The ITC reviewed the ALJ’s decision and upheld it. The ITC agreed with the ALJ’s conclusion that SmartSkin provides a motivation to combine the use of transparent electrodes with a mutual capacitance sensor. The ITC also agreed with the ALJ’s finding that Rekimoto disclosed the limitations in claim 10 that are absent from SmartSkin.
Apple argues that the ITC erred in concluding that SmartSkin in combination with Rekimoto rendered obvious claim 10 of the '607 patent. Apple contends that its design and development story shows that a transparent multitouch screen would not have been obvious to those of skill in the art—Apple’s highly-skilled engineers had [1365]*1365to extensively research and modify the copper mesh SmartSkin design. It asserts that objective evidence reinforces that the '607 patent is not obvious. Specifically, Apple points to evidence that the industry praised the iPhone’s touchscreen; that nearly every major cellphone manufacturer, including Motorola, copied the iPhone’s touchscreen; and that the iPhone was a commercial success.
Apple argues that the ITC improperly employed a hindsight analysis by asking whether the invention was different from the prior art. Second, Apple asserts that the ITC undervalued the ingenuity in measuring capacitance changes and hiding the ITO circuitry, both of which are absent in SmartSkin and Rekimoto. Third, Apple contends that the ITC improperly ignored Apple’s objective evidence.
The ITC and Motorola respond that claim 10 would have been obvious. They contend that claim 10 is not limited to a particular method to measure capacitance and does not require hiding the ITO circuitry to achieve complete transparency. They argue that SmartSkin and Rekimoto disclose every limitation of claim 10. Motorola argues that SmartSkin defines the same problem as the '607 patent—creating a multitouch surface—and provides the solution, including the use of transparent ITO. It points to emails between Apple’s engineers that SmartSkin “could work for multitouch input.”
The ITC and Motorola argue that Apple’s secondary consideration evidence is not adequate to overcome the strong prima facie showing of obviousness. They argue that multiple patents cover the iPhone’s touchscreen and that Apple failed to prove nexus between the '607 patented invention and the commercial success. They contend that the industry praise for the iPhone related to features other than the multitouch screen and assert that Apple presented no evidence of copying.
We are troubled by the ITC’s obviousness analysis. We have repeatedly held that evidence relating to all four Graham factors-—-including objective evidence of secondary considerations—must be considered before determining whether the claimed invention would have been obvious to one of skill in the art at the time of invention. Transocean Offshore Deepwater Drilling, Inc. v. Maersk Drilling USA, Inc., 699 F.3d 1340, 1349 (Fed.Cir.2012); see also Mintz v. Dietz & Watson, Inc., 679 F.3d 1372, 1379 (Fed.Cir.2012) (collecting cases). Indeed, it is axiomatic that “[t]he establishment of a prima facie case ... is not a conclusion on the ultimate issue of obviousness.” Transocean, 699 F.3d at 1348.
The ITC failed to follow this precedent. Prior to even mentioning the secondary considerations, the ALJ concluded that “the evidence clearly and convincingly shows that the '607 patent is obvious in light of SmartSkin in combination with Rekimoto.” J.A. 216. That error warrants vacating the ITC’s decision. The ITC also concluded that claim 10 was obvious and issued its own findings regarding the first three Graham factors (rejecting some of the ALJ conclusions regarding the disclosures in the prior art). The ITC concluded that the '607 patent claims at issue would have been obvious in view of Smartskin in combination with Rekimoto. J.A. 529. The ITC, however, never even mentioned, much less weighed as part of the obviousness analysis, the secondary consideration evidence Apple presented. It stated only that it did not review the ALJ finding regarding secondary considerations. J.A. 523 n. 7. This is not adequate under our law. The ultimate conclusion of obviousness is a legal conclusion to be reached after weighing all the evidence on both sides.3 The ITC analyzed only the disclosure of the prior art references and based solely on that evidence determined [1366]*1366the claims would have been obvious. We conclude that the ITC’s fact findings regarding what the references disclose are supported by substantial evidence. And as the ALJ and the ITC found, the Smartskin reference is very close and expressly recommends as “Conclusions and Directions for Future Work” using transparent ITO electrodes to build a “transparent SmartS-kin sensor.” J.A. 13603. Indeed, the reference teaches that this transparent sensor could be integrated with “most of today’s flat panel displays” because those systems rely on an “active matrix and transparent electrodes.” Id. The ITC erred, however, to the extent that it did not analyze the secondary consideration evidence.
This error was not harmless. Secondary considerations evidence can establish that “an invention appearing to have been obvious in light of the prior art was not” and may be “the most probative and cogent evidence in the record.” Transocean, 699 F.3d at 1349 (quoting Stratoflex, Inc. v. Aeroquip Corp., 713 F.2d 1530, 1538 (Fed.Cir.1983)). This evidence guards against the use of hindsight because it helps “turn back the clock and place the claims in the context that led to their invention.” Mintz, 679 F.3d at 1378. Apple presented compelling secondary considerations evidence that may have rebutted even a strong showing under the first three Graham factors, and the ITC failed to grapple with it.
For example, Apple presented evidence of industry praise by business publications. Time Magazine hailed the iPhone as the 2007 “Invention of the Year” in part because of the phone’s touchscreen and its multitouch capabilities. J.A. 7483-84. Bloomberg Businessweek issued a 2007 article entitled “Apple’s Magic Touch Screen,” in which it labeled the “sophisticated multipoint touch screen” as “[t]he most impressive feature of the new iPhone.” J.A. 7826. Around the same time, Wired Magazine recounted that, after Apple demonstrated the iPhone and its “brilliant screen,” an AT & T executive praised the iPhone as “the best device I have ever seen.” J.A. 8259 (internal quotation marks omitted). The ITC failed to address this evidence and the record does not appear to contain any contrary evidence.
Apple also presented evidence of copying. The ITC failed to address this evidence as well.
Lastly, Apple presented evidence that the iPhone has achieved a high degree of commercial success. Apple presented financial information that showed that the iPhone and related products received billions in dollars of revenue from 2008 to 2010. J.A. 14184. Apple also presented evidence showing a nexus between the undisputed commercial success of the iPhone and the patented multitouch functionality, namely evidence that Apple’s competitors copied its touchscreen and that those in the industry praised the iPhone’s multitouch functionality. The ITC did not address any of this evidence.4
[1367]*1367For the foregoing reasons, we vacate the ITC’s decision that claim 10 of the '607 patent would have been obvious and remand the case for further proceedings. To be clear, we conclude that the ITC fact findings regarding the scope and content of the prior art (what the reference discloses) are supported by substantial evidence. We remand so the ITC can consider that evidence in conjunction with the evidence of secondary considerations and determine in the first instance whether claim 10 would have been obvious to one of skill in the art at the time of the invention.
IV. Noninfringement of '828 Patent
The ALJ construed the term “mathematically fitting an ellipse” to require the method to perform “a mathematical process” whereby “an ellipse is actually fitted to the data,” and, from that ellipse, “various parameters can be calculated.” J.A. 58-70. Based on this construction, the ALJ found that Motorola’s accused products did not infringe because they only measure data from the touchscreen but do not actually fit an ellipse. The ITC declined to review the ALJ’s noninfrinement decision.
Apple argues that the ITC improperly limited the term “mathematically fitting an ellipse” to require calculation of the ellipse parameters after the ellipse has been “actually fitted.” It contends that the specification repeatedly explains that the method fits an ellipse by calculating the parameters of that ellipse or by using default parameters as a baseline—there is no pri- or “fitting” or drawing of the ellipse. Apple asserts that it is irrelevant that the ellipse parameters could, in theory, define other shapes.
The ITC and Motorola contend that the ALJ correctly construed the limitation to require the software to “actually fit[ ]” the ellipse and then calculate the parameters of the ellipse. They contend that the inventors amended the claims during prosecution to overcome a reference that “obtained] measured data ... so long as the measured data happens to be measured from an object that ‘is in general ellipse-like.’ ” J.A. 11920-21. They argue that the plain language of the claim requires the software to “mathematically fit[ ]” an ellipse separate from calculating ellipse parameters. Lastly, they argue that the specification explains that the ellipse parameters are determined by “fitting an ellipse.”
We agree with Apple that the ITC erroneously construed the “mathematically fitting an ellipse” limitation. The plain language requires the software to “mathematically fit[ ]” an ellipse to the data. That process refers to calculating the mathematical parameters that define an ellipse. The dependent claims further support this interpretation. Those claims recite the step of “transmitting one or more ellipse parameters,” '828 patent, claims 2, 3, which implies that the steps in the independent method claim have already calculated the ellipse parameters. Those claims do not imply, as Motorola contends, a separate step of calculating the ellipse parameters.
The remainder of the intrinsic record is in accord with the ordinary meaning of the [1368]*1368claim language. The specification repeatedly explains that the mathematical fitting process creates the parameters of the ellipse. E.g., '828 patent, Fig. 18, col.25 1.54-col.26 1.21. The prosecution history is also consistent with the plain meaning of “mathematically fitting an ellipse.” During prosecution, the applicants distinguished a prior art reference on the basis that the reference obtained data that happened to be “ellipse-like,” ie., the prior art never mathematically fit the received data. J.A. 11920 (emphasis omitted). Those statements are consistent with the ordinary meaning of “mathematically fitting an ellipse” and do not suggest that we should limit the term to require the method to “actually fit[] [an ellipse] to the data.” The correct construction only requires the method to calculate the parameters that define an ellipse. Accordingly, we conclude that the ITC erred in its construction of “mathematically fitting an ellipse.” Having adopted Apple’s construction, we vacate the ITC’s decision that Motorola does not infringe the '828 patent claims and remand the case for further proceedings. We do not accept Motorola’s invitation that we render judgment of noninfringment on appeal. Contrary to Motorola’s arguments, the ITC never found that the Xoom did not infringe under any construction. Nor did Apple concede noninfringment under any construction. See J.A. 133. Apple’s expert did testify that Motorola’s non-Xoom products did not infringe, but that testimony was based on his acceptance of the ITC’s construction of “mathematically fitting an ellipse.” J.A. 30653-55. We thus vacate the ITC’s decision that Motorola does not infringe the '828 patent claims and remand the case to allow the ITC to consider in the first instance whether the accused products infringe under the correct construction of “mathematically fitting an ellipse.”
Conclusion
We have considered the parties’ remaining arguments and find that they are without merit. For the foregoing reasons, we affirm-in-part, reverse-in-part, and vacate-in-part the ITC’s decision and remand for further proceedings.
AFFIRMED-IN-PART, REVERSED-IN-PART, AND VACATED-IN-PART